Eritrez-PubIVle<r
3 NCBl
Page 1 oft
Nucleotide
Search PubMed
bout Entrez
Text Version
Overview
Help | FAQ
Tutorial .^y::j^\VT$7i
New/Noteworthy ■ ; :
PubMed
Journal Browser
MeSH Browser
Single Citation Matcher
Batch Citation Matcher
Clinical Queries
LtnkQut
Cubby
Related Resources;. /.
Order Documents ; i
NLM Gateway v
TOXNET
Consumer Health S -
liiiical Alerts
lihicalTrials.gov
ubMed Central
jpllf
pi: Hum Mutat 1998; 1 l(6):456-60 Related Articles, OMIM, new Books, LinkOut
InteiiS^eficet ]
Assessment of pyrin gene mutations in Turks with familial
Mediterranean fever (FMF).
Chen X, Fischel-Ghodsian N, Cercek A, Hamon M, Ogur G, Lotan R,
j Danon Y, Shohat M.
I Ahmanson Department of Pediatrics, Steven Spielberg Pediatric Research
| Center, Cedars-Sinai Medical Center and UCLA School of Medicine, Los
Angeles, California 90048, USA
Familial Mediterranean fever (FMF) is an autosomal recessive disease
clinically characterized by recurrent short self-limited attacks of fever
accompanied by peritonitis, pleurisy, and arthritis and can lead to
amyloidosis and renal failure in the longer term. It is prevalent mainly in
non-Ashkenazi Jews, Armenians, Turks, and Arabs. Due to the lack of an
accurate diagnostic test, patients often experience years of attacks and
invasive diagnostic procedures before the correct diagnosis is made and
adequate treatment is begun. Recently, the gene responsible for FMF,
denoted pyrin, has been cloned, and three disease mutations have been
described (French FMF Consortium, 1 997 ; International FMF Consortium,
1997). In the current study we assessed the spectrum of mutations in this
I gene in 1 6 unrelated families of Turkish origin. The three previously
reported missense mutations (Met-Ile at codon 680, Met-Val at codon 694,
and Val-Ala at codon 726) accounted for 29 of the 34 disease alleles. In one
patient in whom no disease mutation was identified, the clinical picture was
, atypical enough to raise questions regarding the diagnosis. These results
imply that the origin of FMF in Turkey is heterogeneous, that molecular
diagnosis of FMF is possible in the majority of cases and clinically helpful,
and that delineation of the undiscovered disease mutation(s) in the
remaining cases remains a high priority.
PMID: 9603438 [PubMed - indexed for MEDLINE]
IIMI Abstract
sort ;g i^ialii«feliiiM d i 0rder
http://www .ncbi.nlm.nih.gov/entrez/query .fcgi?cmd=Retrieve&db=PubMed&list_uids=9603... 4/19/02
ConstructiofTofMacromblecular Assemblages In Eukaryotic Processes and flieirTRoIe m Page 1 of 33
Construction of Macromolecular Assemblages in
Eukaryotic Processes and their Role in Human Disease:
Linking RINGs Together
A, Kentsis and K. L. B. Borden*
Department of Physiology & Biophysics, Mount Sinai School of Medicine, New York University, New York,
NY 10029, USA
*Address correspondence to this author at Department of Physiology & Biophysics, Box 1677, Mount Sinai School of
Medicine, New York University, New York, NY 10029; Tel # 212-659-8677; 212-849-2456; e-mail
kathy@inka.mssm.edu
Abstract: Members of the Really Interesting New Gene (RING) family of proteins are found
throughout the cells of eukaryotes and function in processes as diverse as development,
oncogenesis, viral replication and apoptosis. There are over 200 members of the RING family
where membership is based on the presence of a consensus sequence of zinc binding residues.
Outside of these residues there is little sequence homology; however, there are conserved
structural features. Current evidence strongly suggests that RINGs are protein interaction
domains. We examine the features of RING binding motifs in terms of individual cases and the
potential for finding a universal consensus sequence for RING binding domains (FRODOs). This
review examines known and potential functions of RINGs, and attempts to develop a framework
within which their seemingly multivalent cellular roles can be consistently understood in their
structural and biochemical context Interestingly, some RINGs can self-associate as well as bind
other RINGs. The ability to self-associate is typically translated into the annoying propensity of
these domains to aggregate during biochemical characterization. The RINGs of PML, BRCA1,
RAG1, KAPl/TIFlp ) Polycomb proteins, TRAFs and the viral protein Z have been well
characterized in terms of both biochemical studies and functional data and so will serve as focal
points for discussion. We suggest physiological functions for the oligomeric properties of these
domains, such as their role in formation of macromolecular assemblages which function in an
intricate interplay of coupled metal binding, folding and aggregation, and participate in diverse
functions: epigenetic regulation of gene expression, RNA transport, cell cycle control,
ubiquitination, signal transduction and organelle assembly.
INTRODUCTION
By virtue of relating organic and inorganic worlds, organometallic interactions play an
exceedingly important role in living organisms, and undoubtedly present some of the most
fascinating questions in biochemistry. This is particularly true of zinc, which due to its unique
chemical activity and electronic structure is able to facilitate a diverse set of chemical reactions,
http://www.bentham.org/cpps 1 - l/L.B.BordeiVborden_c.htm
4/19/02
"Construction of Macromolecular Assemblages iifEukaryotic Processes and their Role in Page 2 of 33
and gives rise to proteins that constitute one of the largest families in eukaryotes. In spite of the
presence of some proteins that bind zinc in both eu- and archae-bacteriae, these clades lack the
diverse and numerous zinc-binding domains of eukaryotes, which number over 100 in yeast and
more than 500 in Caenorhabditis elegans^ implicating these organometallic macromolecules in
eukaryotic as well as metazoan evolution [1],
Of particular interest to us is a seemingly mysterious group of zinc-binding proteins termed
Really Interesting New Genes (RINGs), with the first protein cloned on the basis of its location
proximal to the MHC regions on chromosome 6. The RING has an unusual arrangement of
cysteine and histidine residues [2] and this coordination consensus (Fig. 1) is similar to other zinc-
binding proteins, namely LIM and PHD domains [3]* RINGs are specifically characterized by
particular core residues in sequences intervening the metal binding sites, as well as their
association with other structural motifs such as B-boxes. RING domains participate in a truly
dazzling array of biochemical processes, including control of protein translation and ubiquitin-
dependent degradation, signal transduction, cell cycle progression and apoptosis, regulation of
transcription and mRNA transport— traits that distinctively define metazoan organisms-as well as
in various human disease states, such as viral infection and cancer, which have also come to
characterize metazoan life.
Fig. (1). Co nsensus sequence of RING and related domains. A, Shows o t her common zinc
bindi ng domains. Parenthetical numbers refer to spacing; dots indicate any residue.
Residues can varyas shown. C8 of LIM can be~C ? H or DTT32], B. The"cbnsensus sequence
tor domains commonly associ^ For
rip: a indicates acidic (D, E); f, aiiphaBc (L, I, V); h, h^opiobk (L7"T, V, M~, Y); o,
rip consehs us sequence was taken fr om [i33]o
Aggregative properties of RING domains have made their biochemical and biophysical
characterization notoriously difficult. However, the paucity of explicit physico-chemical
observations of these proteins should not prevent attempts to understand the physical
underpinnings of their biological behavior. Thus, we examine a body of direct physical
investigations in conjunction with biological studies in an attempt to gain both mechanistic and
holistic insight into the nature of these wide-spread yet elusive entities. In this fashion, we begin
with the exposition of the physico-chemical properties of RING domains, and integrate them in
the context of their biological importance.
Recently, several thorough reviews of structure and function of RING domains have been
written [2-5,134], Thus, we omit discussions of data covered therein, and focus on new
developments. With over two hundred RING proteins reported, we have limited this review to
only a few proteins. In this way, we have endeavored to select examples of this family which reflect
both its rich diversity as well as enable us to define aspects common to all RINGs.
http://www.bentham.org/cpps 1 - 1 /L.B .Borden/borden_c.htm
4/19/02
Construction ofMacromolecular Assemblages in Eukaryotic Processes and their Role in"..." Page 3 of 33
L GENERAL TOPOLOGY
The RING finger is defined by a conserved cysteine-rich consensus sequence (Fig. 1). The
consensus incorporates regions of high and low conservation where in some places spacing
between ligands varies from 9 to 39 residues and in others from 4 to 48. There is a subfamily of
RING fingers termed RING-H2 which have Cys5 substituted with His. There are other cases
where RINGs have a cysteine or histidine substituted with other metal binding residues, such as
Asp or Thr, which are discussed below and in previous reviews [2, 4]. Evolutionary conservation
and differences in loop lengths for the family have been discussed in detail in a previous review
[2]. The spacing in the RING consensus sequence and structures around the second zinc binding
site (see section V, for example) indicate that domain is highly variable. The RING domain
appears to be a stable scaffold with substantial plasticity, forming diverse protein interactions
associated with this large family, as we evaluate below.
To date, three atomic resolution structures of RING domains have been solved: RINGs from
human PML protein and equine herpes protein IEEHV using nuclear magnetic resonance
techniques [6, 7], and the RING from human RAG1 using x-ray diffraction methods [8]. An
extensive comparison of these structures has been reported [9] and will only be summarized here.
The sparcity of structural data is a testament to the biochemical challenge these proteins present,
as well as an impetus for more sophisticated experimental and theoretical treatments. Among the
structures obtained so far, several features appear to be conserved. All three domains bind two
zinc ions per molecule with the inter-metallic distance of 14 A, and use the cross-brace ligation
backbone arrangement [9J. The coordination geometry around zinc-binding site one (Zl) is nearly
identical among the three structures, as is the fold of core residues along the central p-strand (Fig.
2). The conserved core residues for all three structures are just N-terminal to C5
(aromatic/hydrophobic) and the residue C-terminal to C6 (usually a p-branched hydrophobic).
However, there are features that distinguish the three structures. These differences appear to arise
from variable spacing between Zl and Z2. For instance, both IEEHV and RAG1 have a helix
which is absent from PML because the former two proteins have a four residue insert between C6
and C7 relative to PML. The RAG1 structure is missing p-strand 3, which is present in both PML
and IEEHV (Fig. 2). RAG1 is unique in that a classical zinc finger and Zl share a zinc atom
forming a binuclear cluster. Interestingly, despite the interaction with the above zinc finger, site 1
of RAG1 retains its geometric similarity to IEEHV and PML [9]. Pair-wise root mean-square
(RMS) deviation comparisons give the best indication of the structural similarities: PML versus
RAG1 2.1 A (32 out of 43 residues); RAG1 vs. IEEHV 1.9 A (34 from 41 residues); and PML vs.
IEEHV 2.4 A (30 out of 40 residues) [9]. The RMS differences are close to the 2 A RMSD observed
for proteins with -30% sequence identity [10]. Not surprisingly, there is a correlation between the
degree of sequence identity amongst the RINGs and the RMSD (28% between RAG1 and PML,
http://www.bentham.org/cppsl-l/L33orden/borden_c.htm
4/19/02
Construction of Macromolecular Assemblages in Eukaryotic Processes and their Role in ... Page 4 of 33
35% between RAG1 and IEEHV, and 25% between PML and IEEHV). These observations
indicate that RING domains form a conserved structural element and yet possess substantial
structural plasticity, perhaps reflecting the breadth of their biological functions.
II. CHEMICAL PROPERTIES OF RING DOMAINS
We are aware of three RING domains which have been characterized in detail biochemically
and biophysically. Important aspects of chemical properties of RINGs become apparent from this
work, namely the inter-dependence of zinc binding, folding and oligomerization,
A) PML
The PML protein contains a N-terminal RING, and the protein's fusion to the retinoic acid
receptor a is part of the etiology of the majority of acute promyeiocytic leukemia cases. The full
length PML protein forms PML nuclear bodies, multiprotein structures distinct from other
nuclear organelles such as the nucleolus. We studied the 56 residue peptide corresponding to the
PML RING [11]/ Cobalt binding assays indicate that PML binds two cobalt atoms and that the
peptide binds zinc with a higher affinity than cobalt Subsequent analysis of the cobalt spectra by
Roehm and Berg [12] indicate that PML RING binds its first zinc tightly and the second zinc anti-
cooperatively, consistent with observations made for BRCA1 (see part B). Circular dichroism
studies indicate that addition of zinc induced formation of non-helical structure in PML. One-
dimensional l H NMR experiments reveal that zinc addition induces spectral shifts consistent with
formation of a p-sheet and a hydrophobic core, which are reversible upon addition of EDTA,
NMR experiments indicate that this peptide is monomeric but that it aggregates over time [11].
Aggregation is catalyzed by addition of zinc concentrations in excess of PML concentrations [11].
These data emphasize the linkage between zinc binding and folding of RINGs.
B) BRCA1
BRCA1, product of the breast cancer susceptibility gene 1, also has a N-terminal RING
domain, forms microscopically visible bodies in the nucleus, and as PML, participates in
pathogenesis of human disease, contributing to the etiology of breast and ovarian cancers [13].
This domain is subject to some of the familial cancer predisposing mutations where cysteine metal
Hgands are found mutated to glycine [13]. Two groups have examined the metal binding and
biophysical properties of this protein in detail. Using cobalt titration methods, Roehm and Berg
[12] show that BRCA1 binds two zinc atoms per RING, and that metal binding proceeds
sequentially and anticooperatively. The dissociation constant for zinc binding to Zl is 30 nM and
for Z2 500 nM when Zl is saturated. Circular dichroism studies indicate that the apo-BRCAl
contains some elements of secondary structure which are significantly enhanced upon Zl binding,
http://ww.bentham.org/cppsl-l/L33orden/borden_c.htm
4/19/02
"Construction oFMacromolecular Assemblages in Eukaryotic Processes arid their UoTeln .7. Page 5 of 33
and additionally upon Z2 binding. The authors note that cobalt titration spectra reported for the
PML RING [11] show shape changes similar to those obtained for the BRCA1 protein suggesting
that metal binding in PML may also be anticooperative.
Using size exclusion chromatography and chemical cross-linking, Brozvic and colleagues
observe that the majority of the BRCA1 RING domain forms homo-dimers with a small fraction
being in higher order oligomers [14]. Interestingly, there is no evidence of monomers being
present in these studies. Further, homo-dimer formation appears to have a submicromolar JT .
Comparative proteolytic susceptibility studies in combination with mass spectrometry indicate
that the BRCA1 RING domain is much more likely to be cleaved in the absence of zinc,
confirming that zinc is required to form a stable and structured domain. These studies extend to
the C61G mutant found in some familial breast cancers. As expected, only Zl is observed to bind
cobalt. Furthermore, the combination of mass spectrometry and limited proteolysis reveal that the
mutant peptide is more resistant to proteolysis in the presence of metal. The mutant peptide is not
as resistant to proteolytic cleavage as the wild-type peptide in the presence of zinc. The cleavage
pattern indicates that Z2 is less structured than Zl. This Z2 mutant exists as a higher order
oligomer, as observed by size exclusion chromatography.
Fig. (2). Two-dimensional representation of the PML RING structure. Residues CI, C2 , C 5,
an¥C6 consltu^
from the uisconttnuity of coordinating residues beitwec n the two m etal clusters, th e central
|3 2 -strand contains the two conserved hydrophobic core residues and is c ruc ial for the 14 A
inter-zinc distance. Region C-terminal to C8, consisting of a x and b 3 > »s disordered in the
RING of IEEHV [7]. ~ 1
Inferring that the aggregative properties of BRCAl's RING may be due to the absence of some
contextual stabilizing sequences, Brzovic and colleagues have done several studies to determine if
regions outside the RING itself are required for stability. Thus it is suggested that the RING may
not be a stable domain in itself. A comparison of the N-terminal regions of BRCA1 from residues
1-76, 1-112 and 1-172 is carried out using mass spectrometry and limited proteolysis [14]. It is
concluded that residues 1-110 form a unique structural domain. However, because Lys-C and V8
endopeptidases have specific sequence requirements, a limited number of sites in BRCAl's RING
is available for cleavage, making it possible that a cleaved region merely comprises an exposed
flexible loop, and not a veritable unfolded region. A sequence comparison of the C-terminal region
(76-110) postulated to stabilize the RING with appropriate regions C-terminal to the RINGs of
PML, Z and Cbl, indicates that there are two conserved residues: a p-branched hydrophobic
residue and an aromatic residue (Table I). The first of the two residues is present in the peptide
used for the NMR structure determination of PML. The PML RING is stable as a monomer for
several days, but eventually forms a high order aggregate. Thus the presence of one of these
http://www.bentham.org/cpps 1 - l/L.B.Borden/borden_c.htm
4/19/02
Construction of Macromolecular Assemblages in Eukaryotic Processes and their Role in Page 6 of 33
conserved residues is not sufficient for long term stability of PML and does not favor discrete
dimerization of PML unlike the observations for BRCA1. Cbl does not require such additional
sequences for stability as a monomer since a 63 residue construct comprising the RING does not
aggregate over a 6 month period and this construct does not contain the second hydrophobic
conserved residue (see below). Studies with the Z protein, which comprises the entire 90 residue
protein indicate that the additional sequence C-terminal to the RING (residues 72-90) is not
sufficient to inhibit high order aggregation (Kentsis and Borden, unpublished observations). The
data for PML and Cbl indicate that additional sequences are not required for the stability of the
RING in its monomeric state. As of now, there appears to be no general requirement for
additional sequence to impart stability or monomeric behavior to RINGs.
Table I. Comparison of C-Terminal RING Sequences
A RING partner protein for BRCA1 is BARD1 (breast cancer associated RING domain 1).
Yeast 2-hybrid studies show that the RING regions of both proteins are required for specific
heterocomplex formation [15], Both proteins are multidomain proteins which contain N-terminal
RING domains. The association appears to be biologically significant as BRCA1 and BARD1
complexes are detected on some damaged, replicating DNA structures [16]. Biochemical
characterizations including size exclusion chromatography and chemical crosslinking reveal that
like BRCA1, BARD1 homodimerizes [17]. Further analysis indicates that equimolar solutions of
BARD1 (residues 26-152) and BRCA1 (1-112) heterodimerize. Intriguingly, the minimal RING of
BRCA1, comprising residues 1-76, is unable to heterodimerize with BARD1 suggesting that
additional residues are required for the association. Further, BRCA1 (1-76) cannot homodimerize
with longer forms of 1-112 or 1-172 of BRCAL This indicates that residues C-terminal to the
RING are required for efficient association of BRCA1 with itself and with BARD 1. Proteolysis
studies indicate that the minimal stable unit for BARD1 is residues 26-140. Further, these studies
reveal that BRCA1/BARD1 heterodimers are more stable to proteolysis than either set of
homodimers. Together these data suggest that oligomerization is a defining property of RING
biochemistry.
C)RAG1
In depth biophysical and biochemical studies have been carried out by Coleman and co-
workers on the RAG1 RING domain [18]. The RING domain of RAG1 differs from those of PML
and BRCA1 in that its Zl constitutes a binuclear zinc cluster with the adjacent classical zinc
finger ZFA [8], Using proteolytic studies, it was determined that the RING and ZFA form a well
defined structural unit and this construct was used in the studies discussed below. Circular
dichroism studies indicate that introduction of zinc to RING-ZFA results in induction of structure
http : // www. b entham. org/ cpp s 1 - 1 /L. B . B orden/borden_c, htm
4/19/02
Construction of Macromolecular Assemblages Tn Eukaryotic Processes and ffieif Kble in Page 7 of 33
which is lost upon removal of zinc. These studies indicate that 3 zinc atoms are bound per RING-
ZFA complex [18], Substitution of cadmium for zinc results in significantly lower solubility
indicating that the association with zinc is specific. Gel filtration, dynamic light scattering and
velocity sedimentation analyses indicate that the RING-ZFA elutes as a dimer. Further, the apo
largely unfolded form of RING-ZFA aggregates, so dimer formation is dependent on the presence
of zinc. The dissociation constant for dimer formation is approximately 3 and is very similar
to values observed for BRCA1 dimerization. Small angle x-ray scattering experiments indicate
that the RING-ZFA protein is elongated, forming a uniform prolate ellipsoid. Elegant experiments
using RING-ZFA protein fused to the maltose binding protein reveal that the dimer is a parallel
one. Notably, unlike other zinc finger domains such as the glucocorticoid receptor, dimerization is
not driven by association with DNA, rather being dependent on protein-protein interaction [19].
Dynamic light scattering studies with the RING domain alone reveal that this domain aggregates
extensively and does not form discrete dimers. Removal of ZFA exposes hydrophobic residues
resulting in aggregation, illustrating the close relationship among metal binding, folding, and
aggregation of RINGs. The latter appears to be a novel property of RINGs, and even possibly one
of the defining features of their " RINGness."
III. THERMODYNAMICS OF Zn +2 BINDING
The truly outstanding biochemical features of RINGs are imparted by their metallic character.
Zinc binding studies have been reported for the RING domains from COP1, PML, BRCA1 and
RAG1 [11, 12, 18, 20]. Most apo forms of RING domains are largely unstructured and fold in the
presence of zinc, cobalt, or cadmium, although the latter can only support the folding of COP1.
Cadmium may not be a more general ligand because its larger atomic radius may not be easily
incorporated into some zinc binding sites. The tetrahedral coordination of metals is^apparent from
the three-dimensional structures of RINGs [5], and the shape and wavelength maxima observed in
the cobalt absorbance spectra. Roehm and Berg [12] decomposed the cobalt absorbance spectra of
the RING of BRCA1 to arrive at the equilibrium description of metal binding. Site 1, comprised of
four cysteines, has a 50-fold greater affinity of the two metal coordinates, with K of 30 nM [12],
which is expected to be at least an order of magnitude lower for zinc. The two sites exhibit
negative cooperativity, whereby saturation of Zl decreases the affinity of Z2 20-fold. Thus,
kinetically Zl is expected to bind metal first, with Z2 becoming saturated thereafter.
Thermodynamically, in terms of linked equilibria for homotropic binding, this corresponds to a
factor of 20 between the fraction of sites bound under low- and high-saturation conditions [21].
We find it remarkable that a domain with the very non-local topology of a "cross-brace"
should bind zinc with negative cooperativity. The decrease in fractional saturation of Z2 upon Zl
binding in BRCA1 indicates that there exists a regime under which the propensity of site 2 to fold
into a tetrahedral geometry capable of metal coordination is lower than it is in the natively folded
http://www.bentham.org/cppsl- l/L.B.Borden/borden_c.htm
4/19/02
Construction ^MTacrorriolecuIar Assemblages in^EuRarybtic Processes and their Role in ... Page 8 of 33
protein. If the behavior of two sites in BRCA1 is modeled non-cooperatively, the affinity of Z2 is
approximately equal in both the partially folded Zl-bound state (K = 450 nM), as well as the apo-
form (K = 460 nM) which has roughly half the ellipticity of the metal bound state. In this fashion,
d
Zl saturation not only reduces the affinity of Z2 below its value in the native protein, but more
importantly, decreases the propensity of the partially folded chain around Z2 to fold into a metal
binding conformation. In other words, metal binding by site 1 leads to misfolding of site 2, and
implies the formation of a non-native intermediate, which may or may not be off-pathway for the
folding of this RING domain. The existence as well as biological significance of this process are
currently being investigated (Kentsis and Borden, work in progress). These findings raise the
possibility that associations of RINGs with other proteins could be modulated by differential zinc
loading of Z2. Thus, RING recognition could be altered by zinc concentration.
IV. RING ASSOCIATED DOMAINS
Multiple domains, recognized by sequence conservation, are often modularly organized in
proteins. Information about any of these domains can aide in the process of establishing potential
cellular locations, involvement in metabolic pathways or even functions of the proteins which
contain them. Unfortunately, the wide spread use of the RINGs, their presence throughout the cell
and in proteins of diverse functions mean that their presence alone is unable to yield information
regarding the function of the proteins which contain them. Certain protein domains are found
frequently associated with RINGs, some of which have established functions. Often, RINGs with
associated functions tend to have similar associated domains. Associated domains may refine
specificity for potential RING binding domains and their associated functions (see Summary; Fig.
3). The TRAF proteins demonstrate this in particular. TRAFs 2-5 have a N-terminal RING
domain followed by 5 zinc fingers, a coiled coil, and a C-terminal TRAF domain. TRAF1 has all of
these domains with the exception of the RING. Interestingly, some of the RINGs in TRAFs deviate
from the canonical RING consensus in that they have an aspartic acid as the last ligand, which
may be functionally important (see Section XII, part B). The similarity of associated domains and
functions is also clear from the inhibitors of apoptosis (IAP) proteins IAP1, IAP2 and XIAP. These
proteins all have a BIR (baculovirus IAP repeat) domain at the N-terrninus and a C-terminal
RING domain. These proteins bind and inhibit specific caspase reactions [22]. Interestingly, the
anti-apoptotic activity of these proteins is lost when the RING domain is disrupted [23]. Another
intriguing RING arrangement is found in the repair proteins RAD5 and RAD16. In these
proteins, there is an ATPase domain on either side of the RING.
A novel arrangement including two RING domains spaced by an intervening region has been
reported for the Triad and parkin proteins [24, 25]. Triad 1 is a nuclear RING protein
upregulated during granulocytic differentiation in acute leukemia cells [24]. Mutations in the first
RING of the parkin protein are found in patients with autosomal recessive juvenile Parkinson f s
http://www.bentham.org/cpps 1 - 1/L.B .Borden/borden_c.htm
4/19/02
Construction of Macromolecular Assemblages in Eukaryotic Processes and their Role in ... Page 9 of 33
disease [25]. Both of these proteins have an unusual double RING arrangement, where the N-
terminal RING is followed by the "double RING finger linked" (DRIL) or "in between RING
" (IRB) domain, followed by the second RING (Fig. 3). The consensus sequence for this domain is
given in Fig. IB. The RING-DRIL-RING domain is found in other proteins including RBCK2, a
protein kinase C interacting protein. This novel domain is found in 24 proteins, 22 of which have
the domain bracketed by RINGs. Nothing is known about the metal binding or other biochemical
features of this recently reported domain [24, 25].
Most frequently, the RING is associated with cysteine-rich zinc-binding domains known as B-
boxes followed by a leucine coiled coil domain forming the tripartite or RBCC motif [26]. Either
one or two B-boxes is present. The spacing between the RING, B-boxes and the coil is typically
highly conserved, with 38-40 residues between the RING and first B-box, and less than 10 amino
acids between the second B-box and the coil. There is no apparent sequence homology in these
intervening regions. In proteins which contain two B-boxes, these domains are not always
structurally equivalent. For instance, the Bl-box of PML has nascent structure in the absence of
zinc in contrast to the B2-box which requires zinc for structure [27]. The ability of other divalent
cations to induce structural changes differs between the two domains, again highlighting the
potential differences in the domain properties [27]. These differences presumably afford domain
specificity and allow the use of these molecular building blocks in various places throughout the
cell. Members of this class of RING proteins include PML, TIFl a , KAP1, Rfp, RET, XNF-7 and
SSA-Ro (see Section IX). Previous structural studies of the B-box suggest that this domain is of
fundamental importance for alignment of the coiled coil moieties in the RBCC motif [11, 27]. In
support of this, there are BCC proteins, i.e., B-box coil proteins that are missing the RING [26].
The RBCC domain appears to be an integral structural unit requiring all parts for normal
functioning of the proteins in which it is found. Examples of this are found in KAPl/TIFl p, PML,
rfp and MID1&2. Extensive biochemical characterizations of the KAP1 protein by Rauscher and
colleagues indicate that the RBCC is an integral structural unit [28]. A battery of biochemical and
biophysical techniques including gel filtration, analytical ultracentrifugation and analytical gel
electrophoresis on the purified RBCC domain from KAP1 show that the RBCC forms monomers,
trimers and hexamers which appear to co-exist in complex equilibria [28]. Interestingly, mutations
of conserved cysteines to alanines which disrupt the structural integrity of the RING, or mutations
which disrupt coil formation in the RBCC, support the association of mutants with wildtype
KAP1. However, association with full length KAP1 only occurs upon co-expression of KAP1 and
mutant RBCC; simple addition of two purified proteins was not sufficient to reconstitute the
interaction. This is similar to results which show that PML could associate with PML RING
mutants, B-box mutants or small coil deletion mutants only if these proteins are simultaneously
expressed in reticulocyte lysate. It is possible that separately synthesized wildtype and mutant
http://wvvw.bentham.org/cppsl-l/L33orden/borden
4/19/02
Construction of Macromolecular Assemblages in Eukaryotic Processes and their Role i... Page 10 of 33
RBCCs have formed homo-wildtype and homo-mutant oligomers which do not tend to dissociate
and form hetero-oligomers. Separate addition of mutant and wildtype proteins after production in
lysates was not sufficient to reconstitute the interaction between PML and PML RBCC mutants
(K.L.B. Borden, unpublished observations), suggesting that their association is dependent on
folding kinetics and dynamic reorganization of the domains (see Section VII).
Fig. (3). RING a ssociated domains. A schematic of the fre qu ently associated RIN G
doinains. Domain^are defined in the text and c onsensus sequences are given in Fig, b
The RBCC region of KAP1 associates with the KRAB domain of KOX-1 [28]. Interestingly, the
TIFl a RBCC domain, which is highly homologous to KAPl/TIFlp, does not interact with KOX-1
[28]. Mutations in the RBCC which destroy the structural integrity of the RING, B-boxes or
leucine coiled coil result in the inability to bind the KRAB domain. Extensive swapping
experiments indicate that the RING, B-boxes and coiled coil regions are specific to KAP1. For
instance, the RING domain from MIDI is unable to functionally substitute KAPl's RING.
Separate swapping of coiled coils and B-boxes also abolishes the association with KOX-1. The
entire MID-1 RBCC could not reconstitute the interaction with KOX-1. Thus, each component of
the RBCC appears to impart specificity, and the RBCC exists as an intact functional unit.
In certain cases additional domains are present which may further define the function of these
molecules. For instance, in addition to the RBCC domains, both TIFl a and KAP1 contain C-
terminal PHD fingers and bromodomains (Fig. 3). No functional data is known for the PHD
domain/Recent studies indicate that the bromodomain can bind acetylated lysines and therefore
proteins which contain this domain may be involved in chromatin remodeling [29], Consistent
with this function for the bromodomain, both KAP1 and TIFl a are involved in transcriptional
repression whereas other RBCC proteins such as PML and SSA-Ro, which lack the bromodomain
and PHD motif, are not involved in transcriptional repression but appear to function in other
aspects of RNA metabolism (see Section IX).
Several RBCC proteins are associated with a C-terminal B30.2 or rfp domain, named after the
first protein reported with this domain (Fig. 3). Eighteen members of the RBCC family contain
this domain including XNF-7, RPT-1, SSA-Ro and STAF50 [26]. The consensus sequence for this
is given in Fig. IB. The domain corresponds to two immunoglobulin like motifs [30], and it is
speculated that it may be involved in ligand binding but the nature of the Iigand is unknown [26].
The RBCC domain of the RBCK2 protein is very unusual. The domain is comprised of a N-
terminal leucine coiled coil followed by the RING and two B-boxes [31]. It has been postulated
that this protein may actually belong to the RING-IRB/DRIL-RING family of proteins (see
above). In some RBCC proteins, a motif known as NHL is found in the C-terminal region [26, 32].
NHL comes from the identities of the first three proteins found with this domain: NCL, HT2A and
http://www.bentham.org/cppsl-l/L33orden/borden_c.htm
4/19/02
Construction of Macromolecular Assemblages in Eukaryotic Processes and tReirKole i... Page 11 of 33
Lin 41. NCL is involved in regulation of rRNA synthesis; HT2A is a nuclear protein of unknown
function which binds the HIV Tat protein and Lin 41 is involved in cell differentiation. The NHL
motif resembles the WD40 domain, and is thought to be involved in protein protein interactions
(Fig. 3). The motif is typically repeated two to six times and the consensus sequence for it is given
in Fig. IB. Importantly, this protein can be found in proteins with or without the RBCC. The
combination of RING and associated domains suggests that a combination of information in
certain contexts is required for specificity. However, the RING itself can be a determinant in
specificity as not all RINGs can substitute for the actions of others. This is readily apparent in
RING substitution studies. For instance, the RING of BRCA1 binds BAP1 but the RING domain
of rptl does not [33]. Similarly the RING of KAP1 cannot be substituted by the RING of MID2
[28]. Unfortunately, many of the RING associating domains such as DRIL/IRB, NHL and rfp have
no molecular functions associated with them. In other cases, some RINGs such as RAG1 are
associated with classical zinc fingers and these proteins may be involved in forming protein
associations which modulate DNA binding activities of their associated motifs. Once functions
become comprehensively understood for these associated motifs, functional subclasses of RINGs
should be delineated.
V. MOLECULAR RECOGNITION OF RING DOMAINS
Biological functions of RING domains depend on their recognition of other protein motifs.
Here, we describe motifs which have been shown to bind RINGs directly. The RING binding
domains (RBDs) shown here are only for cases where rigorous biochemical analysis indicates a
direct interaction between the RBD and the RING (Table II), Immediately, one can classify RBDs
into two types: domains which bind RINGs through their own RINGs or those which use
heterologous RBDs, some of which we term FRODOs for Funky RING Oligomerization Domains
(see below). Binding of RING domains to other RINGs, as well as to non-RING containing
sequences, has been described for many proteins. It is therefore important to attempt to derive a
sequence-based definition of RING binding. We have applied a phenetic algorithm to derive
similarity scores for sequences of proteins implicated in RING binding and align them using
database-derived scoring matrices such as Blosum and Pam, as implemented in ClustalW 1.7 [34].
Both phenetic comparisons and identity-based scoring matrices indicate that RING domains can
form true homo- and hetero-dimers, whereby particular RING domains bind themselves, such as
the RING of BRCA1, as well as RING domains that are statistically divergent in sequence, such as
BARD1 (data not shown). Because BRCA1/BARD1 heterodimers are more resistant to proteolysis
than either homo-dimer, the recognition between BRCA1 and BARD1 seems to involve folding
and structural reorganization of the RING domains.
Moreover, the specificity of molecular recognition of RINGs bears critical functional
http://^ r ww.bentham.org/cppsl~l/L.B.Borden/borden_c.hto
4/19/02
Constniction^Macromolecular Assemblages in Eukaryotic Processes and their Role i.:. Page"12"of 33
importance. Two illustrative cases are those of recognition of BRCA1 by BAP1 and recognition of
TRAFs by their RING binding partners. BAP1 is a ubiquitin hydrolase which targets BRCA1
and/or its associated proteins for ubiquitin-dependent degradation [33]. This is significant for a
protein such as BRCA1 which plays a role in cell growth and has distinct nuclear and cytoplasmic
functions. Rauscher and colleagues have shown that the RING domain of Rpt-1, a nuclear protein
involved in regulation of interleukin 2 receptor expression among other things [35], cannot
support binding of BRCA1 to BAP1 when substituted for BRCAl's RING domain [33], We
carried out a comparison of sequences comprising RINGs from BRCA1 and Rpt-1. There is more
than 90% sequence identity in the core RING domain constituted by strands 2, 3, and 4; strand 1
also appears to be quite similar, conserving both hydrophobic and charged residue distributions
(data not shown). However, the region immediately C-terminal to site 2 drastically differs between
the two proteins, with Rpt-1 containing an insert (RVPYPFGNLRP) relative to the same sequence
in BRCA1, Similar swapping experiments were carried out with the RING domains of TRAF3
and TRAFS, whereby a construct containing only mid- or C-terminal fusion of the two RING
domains switches the specificity of signaling, activating JNK but not NF- K B, while the wild-type
TRAFS activates both pathways [36]. Moreover, a natural splice variant of TRAF2 exists which
bears a 7-amino acid insertion C-terminal to site 2 in its RING domain, and which is unable to
activate NF- K B in contrast to the normally spliced protein [37]. The importance of the region
around site 2 for the function of RING domains is underscored by the replacement of one of its
Cys by Asp in TRAF3, mimicking the RING of TRAFS which preferentially leads to JNK
activation, and the consequent selective reduction of NF- K B signaling without any effect on JNK
activation [36]. In all, it appears that metal binding by and folding around site 2 are important for
the biological functionality of RINGs, and molecular recognition inherent therein.
Table II. Characteristics of Known RING Binding
Domains
Distillation of the sequence determinants of binding of non-RING containing proteins turned
out to be non-trivial. This is partly due to the low number (eight to date; Table II) of non-RING
proteins that bind RING domains, and even lower number of delineated sequences that bind
RINGs, complicated by the spread of these sequences from plants to man (Table II). In spite of
these confounding variables, it appears that no singular sequence consensus exists for RING-
binding domains. Pseudo-structural methods such as hydrophobic cluster analysis [38] also fail to
reveal any general commonalities (our unpublished observations). However, three proteins which
have been shown to directly bind RINGs through non-RING sequences exhibit a proline-rich
region of high similarity, constituted by the consensus PXBXPJXP, where B and J are Leu/Val
and Ala/Ser, respectively (Table III). Furthermore, all three proteins appear to bind RINGs
http://www.bentham.org/cppsl-l/L33orden/borden_c.htm
4/19/02
Construction of Macromolecular Assemblages in Eukaryotic Processes and their Role C Page i 3 of 33
through the region adjoining Zl, as disruption of Zl of Ste5p leads to loss of binding by Ste4p
[39], mutagenesis of Zl but not of Z2 of MSL2 leads to loss of MSL1 binding [40], and
mutagenesis of Zl in PML and LCMV Z result in no PRH binding [41] whereas regions of the
RINGs of LCMV Z and PML both differ around Z2 (see below), implicating the region around
site 1 as the interface binding to the above proline-rich RBD. Moreover, ATF-1 which has been
shown to bind the RING of BRCA1 does not have any proline-rich sequences like the one above,
and binds BRCAl's RING through the region surrounding Z2 [42]. We postulate that at least two
classes of non-RING RBDs exist, one with the putative consensus PXBXPJXP which binds RING
regions around Zl (FRODO), and another which binds RING regions around Z2 through a non-
proline motif. Comprehensive classification of RBDs is an important future direction for
predicating a classification of RING assemblies.
Table IIL FRODO Consensus Sequence
VI. THERMODYNAMICS OF AGGRE GATION
A biophysical description of RINGs is not fully satisfied by a description of metal binding;
everyone who has worked with RINGs is well aware of their proclivity to aggregate under a wide
variety of conditions, particularly on solid supports commonly used during protein preparation.
Indeed, this is one activity that nearly all RINGs have in common. More importantly, this
behavior appears to be physiologically relevant, as nearly all RING-containing proteins form
large, sub-organelle aggregates with molecular weights in the MDa range. For example, this
property has been well characterized on the cellular level for PML which participates in the
formation of matrix-associated nuclear bodies with a diameter of >1,000 A [43], and a Cys to Ala
point mutation in Zl or Z2 of the RING which abolishes metal binding leads to the disintegration
of PML nuclear bodies [11]. Thus, the linkage and cooperativity of folding and metal binding
inherent in RING thermodynamics are polysteric, taking place during aggregation, as well as
biphasic, involving solid and soluble phases [44], The particular modes by which this aggregation
may occur can be characterized into two general categories. The on-pathway model involves
progressive unidirectional assembly into higher-order structures, while the off-pathway
mechanism depends on a bifurcating assembly from a partially folded state. The latter model has
been postulated as the mechanism of aggregation of a wide variety of proteins (for review see
[45]). It is important to explicitly define the usage of the term aggregation. Although vernacularly
aggregation refers to the non-specific protein oligomerization leading to formation of disordered
assemblages, most physiologically relevant aggregates, such as hemoglobin [46] and Sup35 [47]
are structured, often possessing more secondary structure than the monomeric native state [48].
Thus, by specific aggregation we mean formation of ordered multimers, and not unspecifically
organized flocculent bodies. It is tempting to speculate on the plausibility of the off-pathway model
http://www.bentham.org/cpps 1 - 1 /L.B .Borden7borden__c.htm
4/19/02
Construction of Macromolecular Assemblages in Eukaryotic Processes and theiFRdle 1... " Fage 14 of 33
in light of the apparent misfolding of Z2 upon saturation of site 1 in BRCA1 and requirement of
C-terminal region of BRCAl's RING for homo- and hetero-dimerization. The expression of both
on- and off-pathway modes of aggregation may be state dependent, since at least in the low-order
regime, RING aggregation appears to follow a simple monomer-dimer description, as for example
during the homo- and hetero-dimerization of BRCA1 and BARD1 [14, 17],
VII. BIOLOGICAL SIGNIFICANCE OF RING FOLDING
The above described physico-chemical properties of RING domains are dependent on their
unique cellular milieu. To date, all RINGs have been observed to be intracellular [3, 4] and as
such exist under the reducing conditions of the cytosoi. This prevents the formation of disulfide
bonds, which are usually non-local in character [49], and stabilizing primarily due to the
restriction of available tertiary topologies [50]. An increase in the cooperativity of folding is
associated with the presence of disulfide bonds, the formation of which is linked with folding. In
contrast, the stability of RING domains is assured by coupling folding with metal coordination,
particularly by coordinate residues which are distant from each other in sequence, as they are in
the cross-brace motif of the RING [11]. This cooperativity of metal binding is suggested to be
positive by the complete induction of native structure with the introduction of zinc in the apo-
RING of PML, as determined by NMR [11]. The quantitative determination of the linkage
between folding and metal binding remains to be determined (Kentsis and Borden, work in
progress); the actual stability of RING domains is a product of the intrinsic sequence stability and
the cooperativity of metal binding, ensuring that RING domains are stable under reducing solvent
conditions.
In general, several factors contribute to defining tertiary structure and fold stability including
formation of hydrophobic cores, disulfide formation and metal ligation. Metal coordination is able
to fold stable proteins from inherently unstable sequences, much like disulfide formation, but does
so under reducing conditions characteristic of the intracellular milieu. The unique feature of
metal coordination, and particularly that of zinc, in this role is the relatively fast kinetics of its
ligand exchange, and the consequent potential for highly dynamic protein structures. In contrast
to disulfide bonds, which undergo exchange reactions on the time scale of minutes [51], zinc
exchanges its ligands in milliseconds [52] on the time scale of protein folding [53]. Furthermore,
the completely filled d shell of Zn +2 minimizes ligand field stabilization [54], preventing
thermodynamic barriers to ligand exchange during folding as the metal exchanges its water shell
for amino acid residues. The ability of zinc coordination to stabilize inherently unstable sequences
and maintain thermodynamic stability of proteins under reducing conditions, while facilitating
protein folding dynamics may be important for the physiological behavior of RING domains, such
as misfolding and reorganization of zinc-binding sites and their role in RING aggregation. Both
non-specific and specific aggregation are likely to be in kinetic competition with folding, for which
http://www.bentham.org/cppsl-l/L3.Borden/borden_c.htm
4/19/02
-Construction oTMacromolecular Assemblages in Eulcaryotic Processes and their Role iT.. Page 15 of 33
facile zinc ligand exchange and unhindered folding may be important
VIII. CRITERIA FOR RING BIO-LOGICAL FUNCTION
RING-containing proteins have been implicated in a variety of biological processes, and in light
of the wealth of such data, it is important to define criteria for assigning biological properties to
RING domains. Based on its use of cysteine ligands for zinc binding, and the associated similarity
to classical zinc fingers, upon its discovery the RING was proposed to mediate nucleic acid
binding. Later biochemical studies, including a recent affinity chromatography study of BRCAl's
RING demonstrate that RING domains fail to bind nucleic acids [17, 55]. These results can be
compared with affinity chromatography of DNA-binding classical zinc fingers which are non-
specifically retained by deoxyribose homopolymers [56]. Non-specific binding to nucleic acids has
been observed for the RING domain of the Poly comb protein Psc, but Psc fails to bind to nucleic
acids with any specific affinity [57]. Similarly, Mell8 has been observed to bind DNA-cellulose,
but the protein preparation depended upon enrichment on a heparin column, likewise making it
possible that an indirect interaction contributed to the observed DNA binding [58].
MDM2 and the human homologue HDM2 have been widely cited as RING proteins that bind
nucleic acid through their RING. In addition, the RING domain of MDM2 binds itself and the
highly related MDMX RING in a RING dependent manner [59]. We shall address its putative
nucleic acid binding activity here. In contrast to gel shift assays or affinity chromatography which
can rigorously quantitate degree of binding, RNA-binding by MDM2 was ascertained using a
batch washing method [60]. Nevertheless, MDM2 is observed to bind polyguanidine RNA, and not
double- and single-stranded DNA, and not any other ribose homopolymers. This is an unusual
behavior for a nucleic acid-binding protein, as low affinity binding occurs to varied kinds of
nucleic acids [56]. More crucially, an examination of the relationship between zinc binding and
RNA binding by MDM2 using the same methodology revealed that RNA binding occurs in the
presence of excess EDTA [61], which has been shown to unfold RING domains by zinc chelation
[11, 14]. The independence between zinc binding, e.g., folding, and the proposed RNA binding by
MDM2 indicates that nucleic acid binding is not specific to MDM2's RING domain. The MDM2
family of proteins has a stretch of conserved lysine and arginine residues (KKLKKRNK) between
C6 and C7 which may be able to bind nucleic acids in the absence of a structured protein.
Although unusual, there exist cases of proteins which are active biologically in an unfolded state,
such as synuclein and p-amyloid [5]. We find such a possibility less likely, especially because all
RNA-binding data can be explained by aggregative properties of RING domains. In the work of
Elenbaas and colleagues, sole deletion of the RING domain from MDM2 does not abolish poiy(G)
binding suggesting that the RING is not sufficient for RNA binding. The binding of MDM2 to the
ribosomal protein L5, which comprises the L5/5S rRNA RNP particle could explain this result
[62]. Since cell extracts are used, deletion constructs lacking the L5 binding region can exhibit
http://www.bentham.org/cppsl4/L33order^orden_c.htm
4/19/02
Construction of Macromolecular Assemblages in EuEaryotic Processes and theirTlole T... Page 16 of 33
RNA binding by forming RING-oIigomers with wild-type L5-binding MDM2. Originally we
proposed that the Thr ligand could substitute for Cys at position 3 [63]; see Fig. 1A. Recent
biochemical investigations suggested that Thr is an unlikely zinc ligand and propose a different
coordination scheme. This scheme results in a pattern of conserved cysteine and histidine residues
which is no longer congruent with the constraints of the RING family. Specifically, the spacing
between H4 and C5 is 8 residues, deviating from the consensus of 2-3; similarly, a distance of 7
between C2 and C3 deviates from the consensus of 9-39 [3, 4]. The spacing between H4 and C5 is
particularly problematic because this would greatly alter the dimensions of the conserved central
p-strand which is critical for the formation of the cross-brace motif. This p-strand directly links
Zl and Z2 and is responsible for the invariant 14 A inter-zinc distance of the RING domain. It
remains to be conclusively demonstrated that nucleic acid binding is specifically mediated through
the RING domain of MDM2 by examining the quantitative dependence of RNA-binding on
folding and metal ligation, and even whether MDM2 contains a RING domain through structural
studies. The above work explicitly demonstrates the need for criteria for rigorously assigning
sequences rich in cysteines and histidines as RINGs before assigning biological functions to the
family as a whole.
The above discussion demonstrates that biological behavior of RINGs critically depends on
metal binding and folding into the cross-brace motif, which are in turn dependent on conservation
of metal coordination sites and their spacing, as well as residues critical for fold stability. In the
order of increasing specificity, criteria for assigning biological functions to RINGs:
1) Metal coordination sequence with consensus spacing (Section II). In future, it is possible that
some spacing* may change but in certain regions this spacing could result in such deviation
in terms of 3D structure that the domain in question may not be a RING. From structural
data, it appears that conserved hydrophobic residues N-terminal to C5 and C-terminal to C6
are unlikely to change.
2) Binding of 2 equivalents of zinc per molecule. It seems that zinc binding is anti-cooperative,
with site 1 having higher metal affinity.
3) Induction of folding upon zinc binding, and formation of a functionally competent protein.
4) Cross-brace motif. The cross-brace creates a structural constraint upon the RING, in that
the inter-metal distance is roughly 14 A. For example, although there is significant sequence
similarity between RINGs and LIMs (see Fig. 1), the cross-brace ligation scheme is not used
in the LIM domain.
Proposed biological functions must absolutely depend on the maintenance of these criteria for
their "RINGness."
http://www.bentham.org/cppsl -l/L.B.Border^orden_c.htm
4/19/02
Construction of Macromolecular Assemblages in Eukaryofic Processes and their Role i.„ Page 17 of 33
IX. DIVERSITY OF BIOLOGICAL FUNCTION OF RINGS
Currently, it appears that RING-containing proteins can be grouped into 3 general but
interrelated classes of biological function: those involved in cell growth regulation, those involved
in epigenetic control of cellular differentiation, and those participating in concerted reaction
centers* The function which links these three groups together is the ability of RINGs to form
macromolecular assemblages. The first category is comprised by BRCA1 and BARD1 through
their interactions with pS3, Rb and more downstream cell cycle regulators; PML through its
function in nuclear bodies and cyclin mRNA transport regulation, arenaviral protein Z, and
herpes IEEHV. The second category includes RAG and its role in recombination, Polycomb
proteins in their maintenance of homeotic gene expression, TIFlp/KAPl as a transcriptional
repressor, and MSL and its role in gene dosage compensation. The last category is comprised by
reaction centers involved in ubiquitination of proteins, specifically E3 ubiquitin ligases, and
reaction centers that play a role in signal transduction, such as TRAFs in TNF receptor-mediated
signaling. The unity of RING function in the constitution of concerted reaction centers, as diverse
as E3 ubiquitin ligases and TRAF signaling may appear, is exemplified by the oncoprotein Cbl
which participates in both signaling by growth factors, as well as ubiquitination. Ultimately, by
examining the biology of various RINGs we will integrate the aforementioned biological functions
into their highest archetype.
RING proteins appear to function in formation of large macromolecular assemblages where
the integrity of the RING domain is intimately linked with its ability to form these complexes.
Previously, we have listed those RING domains involved in macromolecular assemblages;
therefore, we will not exhaustively list these here [3, 4]. Below we discuss specifx cases of the
evidence for RING dependent formation of assemblies, the RING interactions necessary to form
these assemblies and their biological properties.
X. CELL CYCLE REGULATION
A) PML
The PML protein is known to form large multiprotein complexes in the nucleus of most cell
types studied. These assemblies, termed PML nuclear bodies, are approximately 1 ^m in diameter
and are heterogeneous in nature. These bodies are disrupted in acute promyelocytic leukemia
(APL), where the PML protein is fused to the retinoic acid receptor alpha (RARA) as a result of a
chromosomal translocation (reviewed in [43]). The resulting PML-RARA fusion protein contains
the RBCC region of PML. Nuclear bodies are disrupted in APL and this contributes to
leukemogenesis. Point mutations which disrupt the RING or B-box structure also result in loss of
nuclear bodies [11, 27]. These point mutations do not disrupt the ability of PML to homodimerize.
http://www.bentham.org/cppsl-l/L33orden/borden_c.htm
4/19/02
Construction of Macromolecular Assemblages ift Eukafyotic Processes ancRheirHoIe i.T Page 18 of 33
This property appears to be driven by the leucine coiled coil region; however, the ability to
dimerize itself is not sufficient for formation of intact nuclear bodies [27]. Mutations in the RING
domain abrogate the formation of these bodies, resulting in PML being uniformly distributed
throughout the nuclear compartment [6, 11]. Many of the biological functions of PML require
both intact nuclear bodies and an intact RING domain. The apoptotic actions of PML [43, 64], its
growth and transformation suppressive actions [65] and its RNA transport properties [66] all
require an intact RING domain.
Mutagenesis studies of the RING were designed to abrogate metal binding and therefore cause
unfolding of the RING. Substitutions of Cys to Ala in either site 1 or site 2 abrogated formation of
nuclear bodies. Thus, two single point mutations in a 69 kDa protein are sufficient to completely
disrupt niacromolecular assemblage formation. Structure based point mutations were made on
the surface of the RING in full length PML, and their ability to disrupt assembly of PML bodies
was assessed using immunofluorescence and confocal laser microscopy [11, 67]. Mutations were
designed to affect the surface charge of the RING without disrupting its ability to fold. Mutation
of charged residues around site 1 results in formation of fewer but extremely large PML nuclear
bodies at the same level of protein expression. Interestingly, mutations in other areas of the
molecule, including regions around site 2, do not have this effect and result in patterns similar to
those observed for wildtype PML. It appears that the region around Zl is important for
niacromolecular assemblage by the PML protein (see Section V) .
Over 15 proteins are known to associate with PML bodies, some of which are incorporated into
the nuclear matrix [43]. These proteins function in a wide range of actions, including chromatin
remodeling, transcriptional regulation, and RNA transport, as well as ribosome function, as
nuclear bodies contain ribosomal components. No DNA or transcriptional machinery is associated
with the bodies making it unlikely that they function directly in transcription. Another RBCC
protein, rfp, associates with PML nuclear bodies [68]. The interaction between PML and rfp is
mediated through the B-box and coil domains of rfp. The RINGs of PML and rfp do not appear to
be required for association. To date, other nuclear body components do not contain RING or
RBCC domains. In fact, other RBCC proteins form nuclear bodies including TIFl a and BRCA1,
both of which are distinct from one another and from PML [16, 69]. Thus nuclear body formation
and the involvement of the RING allows formation of distinct complexes.
PML nuclear bodies are targeted by several viral infections. A small arenaviral RING protein,
Z, can disrupt PML nuclear bodies in viral infection and transfection studies [70]. This 90 residue
protein, binds directly to PML, and translocates the majority of PML and PML nuclear bodies to
the cytoplasm [70]. It is possible that the evolution of an independent viral RING domain is
related to its mimicry of PML. Neither the RING of PML nor of Z is required for this interaction.
The proline rich region N-terminal to the RING of PML interacts directly with the proline rich
http://www.bentham.org/cppsl-l/L.B3order^orden_c.htm 4/19/02
Construction oTMacromolecular Assemblages in Eukaryotic Processes and tHeir Role i... Page 19 of 33
region of Z leaving both RING domains free to associate with other molecules. Interestingly, in
cells expressing PML RING mutants, which normally disrupt nuclear body formation, the Z
protein acts dominantly, reforming nuclear bodies [70]. These data suggest that the RING of Z
can somehow substitute for the RING of PML to allow re-assembly of nuclear bodies. If
constructs with mutations which disrupt the RING of Z are transfected with wildtype PML, the
resulting cytoplasmic and nuclear bodies look identical to cells transfected with wildtype PML and
wildtype Z. Intriguingly, if cells are transfected with RING mutants of both PML and Z,
cytoplasmic bodies form in the absence of apparent nuclear bodies. Thus, there appears to be a
RING docking domain in the nucleus which is recognized by both PML and Z. The presence of
one intact RING domain is sufficient to identify and bind this docking domain.
We have identified a RING binding domain for both PML and Z [41]. The domain is found in
the proline rich region of the proline rich homeodomain protein (PRH). Mutations in site 1 of
either PML or Z destroy the ability of either protein to bind PRH. The consensus sequence around
site 2 varies for all the Z proteins compared to classical RING fingers like PML. Thus it is unlikely
that PML and Z would recognize similar proteins through regions surrounding site 2. Instead, it
appears that the PRH protein recognizes the region around site 1. This RING binding domain is
unusual in that it recognizes two distinct RINGs. Further, PRH associates with a subset of PML
bodies in APL patient-derived cells as well as with bodies of unknown function. The composition
of the second type of bodies is currently unknown but may well contain additional RING proteins.
The fact that PRH recognizes the most structurally and sequentially conserved regions of these
two RINGs, namely site 1, suggests that PRH may bind other RINGs, and possesses some features
for universal RING recognition.
Our results suggest that PML nuclear bodies are involved in RNA transport and translation.
PML and Z associate with translation factors including the ribosomal P proteins and eIF-4E [9,
66, 71]. The association with eIF-4E is RING independent [66, 71]. However, in vitro, the RBCC
region of PML and the Z protein can inhibit translation with no decrease in RNA production and
these actions require the RING [66, 71]. The association of both proteins with several different
translation components suggests that these proteins can actively interfere with pre-translation
complexes or effectively form translational repressor complexes. In cell culture, PML inhibits
transport of selected RNAs in a RING dependent manner [66]. The precise molecular mechanism
of PML controlled RNA transport is not established but it is clear that it requires the RING
domain, most likely for proper nuclear body assembly.
B) BRCA1
Germline mutations in BRCA1 account for nearly 50% of familial breast cancers [13]. BRCA1
is 1863 amino acids but has few sequence motifs suggestive of function. In the N- terminus there is
http://www.bentham.org/cppsl -l/L.B.Borden/borden_c. htm 4/19/02
Construction ofMacromblecular Assemblages inliukaryotic Processes and their Role i7.. Page 20 of 33
the RING and in the C-terminns there is an acidic region followed by two copies of a novel motif,
BRCT. BRCA1 is found in the nucleus and cytoplasm in intact cells [72, 73] where some isoforms
of BRCA1 do not have a nuclear localization signal. BRCA1 protein co-fractionates with RNA
Polymerase II and in a RING dependent manner with E2F, ATF, cyclins and cdk2 [42, 74], E2F
and ATF function directly in transcription. The C-terminus of BRCA1 is capable of activating
transcription [75]. BRCA1 can modulate transcription of ATF; however, the C61G familial
mutation in the RING of BRCA1 no longer has this activity [42]. Taken together, these data
suggest that BRCA1 functions in transcriptional regulation, but the involvement of the RING
domain in this process is unclear at this time.
The BRCA1 protein forms large nuclear structures in a cell cycle dependent manner [16].
BRCA1 and BARD1 localize to discrete foci during S phase and disperse during other stages of
the cell cycle [16]. BRCA1 and BARD1 complexes have been detected on damaged DNA [16].
BRCA1 physically associates with the recombination/repair protein RAD51 [16]. BRCA1 and
RAD 51 co-localize on synaptonemal meiotic chromosomes suggesting that the BRCA1 complex
could play roles in DNA replication, cell-cycle regulation and maintenance of genomic integrity.
Association of the ubiquitin hydrolase molecule BAP1 with BRCA1 may indicate a mechanism for
control of the function of these complexes by degradation or by re-targeting BRCA1 complexes to
other parts of the cell [33]. For instance, ubiquitin-like modification of Ran-GAP causes the
molecule to move from the cytoplasm to the nuclear envelope and does not cause degradation of
RanGAP [76]. All in all, although the RING dependence of BRCAl's cellular functions remains
unclear, its RING domain seems important for association with other proteins and organization of
macromolecular protein assemblages which are involved in regulation of cell growth.
XI. EPIGENETIC REGULATION OF DIFFERENTIATION
A) MSL
MSL proteins were first identified as participants in gene dosage compensation in Drosophila^
upregulating the expression of X-linked genes in male flies. They exist in a 1 MDa complex as
observed by gel filtration [77], which requires all members, MSL1 through 3, for proper
localization to the X chromosome. MSL2 contains the RING domain which is required for its
interaction with MSL1, and Leu to Pro substitutions near Zl abolish binding, reminiscent of
BRCA1-BARD1 interaction, implicating the region around Zl in RBD binding (see Section V).
The functional importance of Z2 is evident from stable expression of Z2 mutants of MSL2 in yeast
but their absence in Drosophila [40]. In general, the importance of the RING domain for proper
male fly development is illustrated by the lethality of mutating the zinc coordination cluster in
MSL2 which leads to the disruption of MSL complex formation and lack of localization to X
chromosomes [78]. MSL1, the RING-binding partner of MSL2, has a N-terminus characteristic of
http://www.bentharn.org/cpps 1 -l/L.B.Borden/borden_c.htm 4/1 9/02
Construction of Macromolecular Assemblages in Eukaryotic Processes and their Role i... Page 21 of 33
factors involved in chromatin remodeling [79], and it appears that the MSL complex with its
dependence on aggregative RING domain function is an epigenetic regulator of gene expression.
B) TIFlp/KAPl
The KRAB domain is one of the most numerous DNA-dependent transcriptional repression
domains [80]. TIFlp/KAPl was initially identified as a co-repressor that binds to the KRAB
domain [81]. TIFlp/KAPl contains a RING domain within a RBCC motif, as well as a histone
acetyltransferase bromodomain and a PHD finger. Like other RING containing proteins,
TIFlp/KAPl is observed by bandshift assays to be in 1+ MDa complexes in mammalian nuclear
extracts [82]. In vitro, the RING domain is required for association [28], and in vivo, the RING
domain significantly potentiates the interaction with KRAB, as well as its repression of
transcription [83]. TIFlp/KAPl binds HP family of proteins, which exert dose-dependent effects
on heterochromatin-mediated gene silencing [84]. Although no RING mutagenesis data exists to
determine the dependence of epigenetic repression by TIFlp/KAPl on its RING, observation of
TIFlp/KAPl in a ternary complex with DNA-bound KRAB domain [81], a granular pattern of
immunofluorescence in the nucleus [82], coupled with the above observations strongly implicates
TIFlp/KAPl and its RING in the assembly of epigenetic macromolecular complexes that play a
role in transcriptional repression by way of chromatin remodeling. Delineation of the precise
function of the RING domain remains to be done, as well as determination if TIFlp/KAPl is a
scaffold for the interplay between eu- and hetero-chromatin remodeling or the modulator of
chromatin structure itself. It has been suggested that TIFlp/KAPl -associated chromatin
remodeling factor HP1/MOD plays a role in modulating higher order chromatin structure [85]. It
would not be surprising if TIFlp/KAPl plays a role in this process by forming a higher order
macromolecular scaffolds for high order chromatin remodeling proteins.
C) RLIM
RLIM is a RING-H2 protein which binds LHX2 [86]. The latter contains a LEVI domain which
is required for inhibitory effects on homeodomain transcription factors and for transcriptional
activation. RLIM acts as a negative co-regulator of transcription by recruiting the histone
deacetylases HDAC2 and Sin3A to the LIM domain. The LIM domain of LHX2 is necessary and
sufficient for association with RLIM. It appears that RING-H2 domains are not uniquely
distinguished from RING domains with respect to their biological function, as both RING and
RING-H2 containing proteins are involved in epigenetic transcriptional regulation and ubiquitin-
dependent protein degradation (see below).
D) Polycomb
The Polycomb proteins constitute the most explicit involvement of RING domains in epigenetic
http://www.bentham.org/cppsl-l/L.B.BordeE^borden_c.htm
4/19/02
Construction of Macromolecular Assemblages in Eukaryotic iProcesses and their Role i... Page 22 of 33
regulation of gene expression. The Polycomb group of proteins is important for the maintenance
of repressed state of homeotic gene expression in cells where homeotic gene was originally inactive
[87]. Bmil is a RING-containing mouse protein homologous to Drosophila Posterior Sex Combs
(Psc) and Suppressor two ofzeste (Su(Z)2) [88J. Bmil exists in a 2-5 MDa complex, and mutation of
zinc-ligating Cys to Phe causes anterior-posterior transformations of vertebrae with a nuclear
diffuse pattern [89], indicating that both the assembly of Polycomb complexes, as well as their
homeotic function are RING-dependent. Moreover, Bmil forms a RING- RING dimer with
dinG/RINGlb, which is disrupted by the said Cys mutation in Bmil [88]. Since transcriptional
repression by Polycomb proteins [90] is mediated by alteration of chromatin structure [91], and
the RING domain is important for their function, it seems that Polycomb proteins are
macromolecular assemblage epigenetic homeotic regulators.
XII. CONCERTED REACTION CENTERS
A) Ubiquitin Targeting Centers: APC, SCF, VHL
E3 ubiquitin ligases are constituted by 4 classes (Ubrl, HECT, SCF, and APC), all of which
contain RING proteins, and possess similar architectures [92]. The anaphase-promoting complex
(APC) also known as the cyclosome, functions during cell cycle progression to degrade cyclins as
well as a number of other regulatory proteins [93]. APC11, also known as Cdc53 [94], contains a
RING domain and potently activates autoubiquitination of Cdc34 E2 ubiquitin-conjugating
enzyme, constituting a minimal ubiquitin ligase [94]. The RING domains in E3 ubiquitin ligases
are unique in that site 2 contains HH ligands instead of the classical RING coordinates CH. It has
been proposed that these RING-H2 domains are specific for ubiquitin ligase targeting.
Observation of ubiquitination activity of proteins which contains classical RING domains (see
below) undermines the uniqueness of RING-H2 domains in E3 ligase function. Further, RING-H2
proteins can function outside of the ubiquitin pathway (see Section XI, part C; and below).
APC complexes control entry into anaphase and exit from mitosis in budding yeast [94]. These
complexes are found in higher organisms and are also required for regulation of the cell cycle [92].
The enzymatic pathway for ubiquitin conjugation and modification is classically considered as
follows: activation of ubiquitin with formation of enzyme-ubiquitin thioester by El activating
enzyme, transfer of ubiquitin thioester to E2 conjugating enzyme, and promotion of ubiquitin onto
target protein by the E3 ligase, or E2 enzyme in some cases. The substrate recognition step and
subsequent ubiquitin ligation are typically mediated by E3 ligases. Three E3 ubiquitin ligase
complexes: APC, SCF (Skpl/Cdc53/F-box) and VCB (Von Hippel, Elongin BC) contain RING-H2
proteins in their assemblies [92, 94, 95]. The E3 ligase complexes can be subdivided into three
distinct functional domains: the F-box subunit recognizes substrates through specific protein
interaction domains, the Skpl subunit links the F-box to the remaining subunits and the
http://www.bentham.org/cpps 1 -l/L,B.Borden/borden_c.htm
4/19/02
Construction of Macromolecular Assemblages in Eukaryotic Processes and their Role L. Page 23 of 33
cullins/Cdc53/Hrtl subunits recruit the E2/Cdc34 ubiquitin conjugating enzyme and comprise the
ubiquitin ligase core. The F-box has been proposed to mediate specific targeting of the ligase to
proteins to be degraded and the RING domain is necessary for this function. The basic ubiquitin
ligase complexes can be assembled from the RING-H2/cuIIin module. Thus, RINGs are part of
complexes involved in catalyzing polyubiquitination and as such contribute to the function of E3
ligases. These complexes can interact with different adaptor proteins, thereby recruiting different
binding partners through the F-box, and the RING is important for this specificity.
In the SCF complex, RING-H2 proteins referred to as ROC1 and ROC2 associate directly with
cullins 1-5 [96]. The APC homologue to ROC1 is APC11; however, APC11 does not bind cullins 1-
4 but binds a yeast homologue APC2 [96]. Conversely, ROC1 cannot bind APC2. Although there
is extensive sequence homology between ROCs and APC proteins, their heterologous interactions
are distinct. For both complexes, the RING components, ROC1 and APC11 are required for
ubiquitin ligase activity of the complex/Interestingly, mutations in site 1 of ROCl's RING reduce
ubiquitin ligase activity whereas mutations in site 2 destroy them altogether [96]. Thus, protein
interactions made by site 2 appear to be critical for normal functioning of this complex (see
Section III and V). In a recent discovery, the VHL protein, which is mutated in most kidney
cancers, was found to be part of an E3 ubiquitin ligase complex [95]. Here, the RING-H2
homologue is referred to as Rbx (for RING box). Rbx directly binds Cdc53/cullin, Cdc34, but not
Skp, again emphasizing the specificity of RING complex formation. The RING-H2 domains
involved in these complexes are small proteins, on the order of 100 residues and do not appear to
have other motifs. Furthermore, the only common features among these complexes are the RING-
ER protein and its associated cullin. For SCF, APC and VCB E3 ligase complexes, the RING-H2
component is required for the formation of an active complex, presumably by mediating crucial
protein interactions in the construction of a macromolecular concerted reaction center.
There are other RING proteins which are involved in various aspects of protein degradation,
including transmembrane protein transport, which is tightly coupled to both protein synthesis and
protein degradation. The Ubrl protein functions as a ubiquitin ligase similar to the above RINGs.
Ubrl contains a RING-H2 finger and classical zinc finger protein [97, 98]. The putative RING
protein, MDM2, is involved in ubiquitin dependent degradation of p53 [99]. Cbl, which has a
typical RING domain, is involved in the ubiquitination of growth factor receptors through its
RING domain [100]. The ubiquitin conjugating enzyme UbcM4, which is necessary for mouse
development, interacts with a family of UbcM4 interacting proteins (UIPs) which belong to the
RING finger family [101]. UIP48 and UIP28 are RBCC proteins and may recruit UbcM4 to SCF-
like complexes. The Der3p protein located in the ER membrane has 5 transmembrane regions and
a C-terminal RING-H2 domain located in the lumen of the ER. The RING-H2 domain is required
for correct localization of Der3p and for the ability of Der3p to degrade specific proteins [102].
http ://www.bentham. org/cpps 1 - 1 /LB .Borden/borden__c.htm
4/19/02
Construction of Macromolecular Assemblages in Eukaryotic Processes and their Role i... Page 24 of 33
Der3p appears to act prior to retrograde transport of ER membrane and lumenal proteins to the
cytoplasm for degradation via the ubiquitin-proteosome system. Neurodapl is associated with the
cytosolic face of the ER and Golgi and with the postsynaptic region of the cytoplasmic membrane
[103].
Recent studies utilizing purified RING proteins indicate that in vitro some RING proteins are
required for the action of E3 ubiquitin ligases. A recent report from Lorick and colleagues [104]
indicate that a new member of the RING-H2 family, A07, binds directly to E2. GST-A07 is able
to associate with purified preparations of the E2 ligase, UbcHSB, and undergo ubiquitination. The
RING is required for the association with UbcHSB but not for ubiquitination itself. Interestingly,
the A07 protein is not modified in these studies, while GST is ubiquitinated. These studies showed
that association of A07 through its RING is necessary for association with UbcHSB but is not
sufficient for ubiquitination. Lorick and colleagues extend this work to other RING proteins and
show that as GST fusions Prajal, NF-X1, kf-1, TRC8, Siah-1 and BRCA1 can all bind UbcHSB
and support ubiquitination. In all cases an intact RING is required for association with UbcHSB
and treatment of proteins with zinc chelators destroys this activity as do mutations of conserved
zinc ligating residues. Similar studies done independently show that the RING of Cbl acts
similarly to the above proteins [105]. Lorick and colleagues [104] suggest that it is unlikely that
RINGs bind E2 enzymes and form catalytic thiol ester intermediates with ubiquitin. Furthermore,
it is unlikely that the RING provides a site of interaction with E2 for direct ubiquitin transfer
from E2 to an available target lysine. They favor the hypothesis that the RING and surrounding
regions associate with E2-ubiqutin and provide a favorable environment for the transfer of the
ubiquitin from E2 to the targeted lysine.
These results raise several intriguing questions related to RING function. In the above in vitro
studies, RING domains from diverse proteins are able to substitute for the ubiquitination function
of the A07 RING protein. This plasticity is surprising given that RINGs typically cannot
substitute for each other with regard to other biological functions of RINGs in vivo and in vitro,
such as binding of BAP1 by BRCA1, and differential signal transduction by TRAFs (see Section V
and Section XII, Part B). In ubiquitination assays which demonstrate the exchangeability of
RINGs, transfer of ubiquitin occurs non-specifically onto the RING proteins themselves or GST.
Notably, the RING itself never appears to be ubiquitinated. In spite of the fact that a subset of
RINGs have been shown to be able to catalyze ubiquitin transfer reactions, it is unlikely that this
is the defining function of RING domains. It is important to note that RINGs of PML and MAT1
are unable to support catalysis of ubiquitin transfer by UbcHSB or Cdc34 (Borden, Chen and Pan,
unpublished observations). It is clear that some RINGs participate in ubiquitin transfer reactions,
as evident by physiologic observations [92] as well as the aforementioned in vitro assays. It is the
latter, however, which demonstrate the interchangeability of RINGs in this function, but do so in a
http://www.bentham.org/cppsl-l/L33orderi/borden_c.htm
4/19/02
Construction oFMacromolecular Assemblages in Eukaryotic Processes and their Role i... Page 25 of 33
non-physiologic manner, lacking specificity demonstrated by the transfer of ubiquitin even onto
GST [1041, which is presumably merely positioned in spatial proximity to the ubiquitin ligase. In
general, such non-physiologic behavior can be due to two reasons: 1) substituted RINGs are
physiologically irrelevant and as such do not confer specificity, or 2) undefined factor(s) are
absent from the assay which would make ubiquitin transfer target specific. In either scenario v the
function of diverse RINGs in supporting albeit non-specific ubiquitin transfer emphasizes the
more universal and characteristic function of RINGs as molecular building blocks providing a
scaffold for this and other processes. The fact that RINGs can participate in ubiquitin transfer to
other proteins as well as themselves could represent a novel auto-regulatory mechanism, whereby
it may target itself for ubiquitin-dependent degradation or for modification of biological activity
such as intracellular targeting that occurs with modification by SUMO-1 of RanGAP [106].
E3 ubiquitin iigases exist as large macromoiecular assemblies [94], which critically depend
upon the RING domain for their organization and function. RING-containing proteins specific to
E3 Iigases are on the order of 100 residues containing no conventional sequence motifs other than
RINGs. We propose that the thioester-ubiquitin transfer needs spatial coupling of E2 and E3
catalytic systems and their targets, requiring a surface or scaffold provided by the oligomerization
of the RING domain, and even possibly by the directive recruitment of proteins to be degraded by
the RING in conjunction with the F-box.
B) Signaling Centers: TRAFs
TRAFs are tumor necrosis factor-receptor (TNFR) associated proteins which constitute an
upstream component of one of the TNFR signaling pathways; the other consisting of signaling
through the death domain (DD) of TNFRs [107]. All TRAFs except TRAF1 contain a RING
domain, interact with DD-lacking TNFR members, and TRAF constructs lacking RING domains
behave as dominant negative inhibitors of TNFR signaling, such as NF- K B activation [108]. CAP-
1, a homologue of TRAF2, binds CD40, and RING deletion abrogates this binding [109], as well as
Ig class switching in a B cell line [110]. Similar behavior has been observed for TRAF3 [111], and
TRAFS [112]. These signaling aggregates are insoluble in non-denaturing buffers, not only
oligomerizing TNF receptors, which is important for their activation, but also recruiting
downstream kinases and phosphatases, such as MAPK/MEKK1 by TRAF2; this association being
dependent on the RING domain [113]. Furthermore, many RING-H2 proteins are actually found
in membrane proteins and appear to be involved in receptor oligomerization directly. For
instance, the RING-H2 protein rapsyn is associated with the plasma membrane and is involved in
aggregation and clustering of the acetylcholine receptor [114]. Moreover, RING-dependent
oligomerization stimulates the catalytic activity of MAPK/MEKK1 and its autophosphorylation,
necessary for JNK1 activation. The arrangement of TRAFs is reminiscent of PML, since both
contain a RING and a leucine coiled-coil, and the coiled-coil is important for self-oligomerization
http://www.bentham.org/cppsl -l/L.B3orden/borden_c.htm
4/19/02
Construction of Macromolecular Assemblages in Eukaryotic Processes and their Role i... Page 26 of 33
of TRAFs [115]. More importantly, the RING domain is not only necessary for transduction along
the TNFR pathway, but also appears to impart specificity to signaling. Overexpression of TRAFS
leads to NF- K B activation and is dependent only on lymphotoxin p receptor activation, not
interacting with TNFR2; similarly for TRAF2 and TRAF3 [116, 117]. On the other hand, TRAF3
overexpression inhibits NF~ K B signaling, and is dependent on TNFR2 and CD40 activation [112].
Interestingly, a C8 to Asp mutation in Z2 coordination of TRAF3's RING, mimicking the RING of
TRAFS, selectively reduces NF- K B signaling, without affecting JNK activation [36]. A construct
containing only a mid- or C-terminal fusion of the RING domains of TRAF3 and TRAFS activates
JNK, but not NF- K B. This implicates the Z2 region of the RING in TRAF signaling (see Section III
and IV). Interestingly, there exists a natural splice variant of TRAF2, which is homologous with
TRAFS, bearing a 7-amino acid insertion in the Z2 region of the RING, which in contrast to
TRAF2 does not activate NF- K B signaling [37]. Thus, it seems that the RING domains of TRAFs
are important for protein scaffolding and formation of macromolecular signaling centers, and
even possibly for endowing these assemblages with signaling specificity by way of variable RING
domains.
C) Organelle Assembly
Several RING proteins have been implicated in vesicular transport and peroxisomal assembly.
Vacl, a yeast RING protein, is implicated in regulation of vesicle docking and fusion [118]. The
Neurodapl protein mentioned above is involved in sorting of presynaptic proteins and vesicles
[103]. The RBCC protein, AUDI, contains a GTPase function and plays a role in vesicular
trafficking [119]. Several RING proteins are involved in peroxisomal biogenesis, and their
disruption contributes to developmental disorders of peroxisome genesis [120]. Truncation of
Paflp, an integral membrane protein of peroxisomes, leads to Zellweger's syndrome [121].
Mutations in the RING of other peroxisomal proteins result in a lack of peroxisomal structures,
e.g. Per8p or Pas7p [122-124]. A novel RBCC protein BERT interacts with class V myosins and
may cooperate with myosin V during cellular outgrowth and organelle transport [125]. Although
the exact role of RINGs in organelle assembly is only beginning to be appreciated, their
involvement in vesicular sorting, organelle transport, and remodeling of cell morphology suggests
that they may play a role in the constitution of the highest macromolecular concerted reaction
centers, those that build and remodel cells.
D) Cbl
The Cbl protein contains a number of identifiable sequence motifs, such as a PTB domain and
a SH3 binding domain [126]. Positioned in the middle of this 120 kDa cytoplasmic protein is a
RING domain. This is an unusual position for the RING domain which is usually N- or C-
terminal (see Section IV) [4]. Oncogenic variants of Cbl contain mutated forms of the RING [127,
http://ww.bentham.org/cppsl-l/L33order^orden_c.htm
4/19/02
Contraction of Macromolecular Assemblages in Eukaryotic Processes and their Role i... Page 27 of 33
128], We have carried out NMR studies on a 63 residue protein which corresponds to the RING
domain of Cbl. Our data indicate that this RING is stable over months. ID l U NMR spectra
recorded immediately after protein purification are shown in Fig. 4A. Below, a spectrum recorded
after 6 months in solution is shown (Fig. 4B). These spectra show little change in the intervening 6
months nor is there any sign of significant precipitation. In contrast, the 56 residue PML RING
aggregates and precipitates after 48 hours in solution [11]. These data clearly indicate that
minimal RING domains can be monomerically stable and can maintain a folded conformation.
Stability of RING domains awaits its full examination (see Section II, part B).
Fig. (4). ID l U NMR spectra of a 63 residue protein correspond
1 1.7 Tesia Bruker
the day after protein preparation Id panel B was coflected 6 months later. Data cbiiection
parameters were nearly identical for the two samples with the same number of scans were
collected in both cases! Note that the spectra show little change over time. [
The relationship of the above described behavior of Cbl's RING and its cellular function is
unclear. However, Cbl is unique in the biological repertoire of RING domains in that it
participates in growth factor receptor signaling, for example by EGFR [129] and PGDFR [130], as
well as in ubiquitin targeting of these receptors and their downregulation, and in this fashion
behaves as a proto-oneogene. This downregulation is important for signaling by these receptors, as
Cbl is crucial for desensitization of EGFR signaling [100]. The nexus of signaling and protein
degradation, presumably united by Cbl's participation in an E3 ubiquitin ligase [105],
demonstrates the interwoven nature of cellular biochemistry, whereby a single protein domain
with a finite repertoire of chemical and physical properties (Sections II-VII) acts in a diverse set of
cellular processes, spanning epigenesis and cell cycle regulation, protein degradation and
signaling.
SUMMARY
By virtue of being organometallic proteins and participating in most processes which we
associate with eukaryotic life, RINGs exemplify the symbiosis of inorganic chemistry, structural
biology, chemistry and physics, and the biological role of these processes. Such a holistic
understanding of RING biology is particularly important for making sense of the importance of
RINGs, which once separated into their genetic, structural, and physical component parts fail to
articulate why they are indeed so important for living things, and why their disruption leads to
disease.
We have sought to identify criteria which define RINGs structurally, biochemically and
functionally, inter-relating these levels of analysis in order to distill an understanding of RINGs
which unifies their seemingly diverse biological behavior. Due to their zinc-binding properties,
http://www.bentham.org/cpps 1 - 1 /L.B .Borden/borden_c.htm
4/19/02
Construction of Macromolecular Assemblages in Eukaryotic Processes and their Role i... Page 28 of 33
RINGs have been asserted to bind nucleic acids, a contention we show to be unlikely. Instead,
RINGs appear to participate in a variety of protein-protein interactions, both oligomerically and
heterologously, on their own, as well as in the context of other associated protein motifs. The
RING domain is defined by a particular spacing of zinc-coordinating residues, which it shares
with the LIM domain, although the two can be differentiated based on hydrophobic core residues
of RINGs and their use of a non-local cross-brace arrangement to bind zinc. The structure around
metal site 1 is quite conserved, with regions around site 2 exhibiting substantial variability. This
plasticity may endow RING oligomers with specificity, whereby distinct complexes are formed by
different RINGs, Oligomeric properties of RINGs are both low- and high-order, leading to the
formation of dimers as well as large MDa aggregates. The processes of aggregation, folding, and
zinc binding appear to be intimately linked, and this linkage is biologically crucial. Folding into
the cross-brace conformation and metal binding are cooperative, with site 1 having higher affinity
than site 2, but binding between two metal clusters is anticooperative, whereby binding by site 1
leads to misfolding of regions around site 2. Since zinc ligand exchange occurs on the timescale of
protein folding, these properties may be important for RING-RING interactions. We have
identified heterologous RBDs, termed FRODOs, which appear to bind RINGs through regions
around site 1, utilizing a proiine-rich motif. The diversity of RINGs suggests that there will be
other classes of RING binding domains. The monomeric stability of the RING domain has been
debated. We show that RINGs can be monomerically stable in the absence of other sequences, but
are not necessarily so, illustrating the aforementioned equilibrium among metal binding, folding
and aggregation.
Indeed, what seems to unite all RINGs in their diverse biological functions is this ability to
form macromolecular complexes. Such assemblages are important for the function of epigenetic
regulation of gene expression, cell cycle regulation, ubiquitin-dependent proteolysis, and signal
transduction. RINGs provide reaction surfaces and molecular scaffolds, both in a classical
chemical sense of thioester-ubiquitin ligases and in a complex sense of epigenetic gene expression
regulators which operate with chromosomal substrates. In both cases, metal binding, folding and
aggregation of RING domains are necessary for the function of these assemblages. It is also
possible that in addition to this structural role, RINGs may have directive functions, especially in
light of the diversity and distinctiveness of RING assemblages, but at the moment this is unclear.
Even in the purely architectural role, the construction of macromolecular assemblages by RINGs
has important implications for eukaryotic cells, which must carry out their complex functions in
the crowded cellular milieu. Scaffolding for simple chemical reactions and complex biological
processes is critical in that it provides spatial and temporal coupling for cellular functions such as
protein degradation, chromatin remodeling and signal transduction. Aberrant organization of
these assemblages plays a role in oncogenesis and viral infection, as well as other disorders like
Zellweger's syndrome. A better understanding of RINGs inherently involves a deeper
http://www.bentham.org/cppsl - 1/L.B 3orden/borden_c.htm 4/1 9/02
Construction of Macromolecular Assemblages in Eukaryotic Processes and their Role!... Page 29 of 33
understanding of organometallic interactions, folding and aggregation, intermolecular recognition
and protein design. This knowledge will lead to both potential therapies for diseases of RING
dysfunction, as well as a more profound understanding of cellular processes that define
eukaryotes, which occur within the complex nexus of eukaryotic life and their crowded cellular
milieu, both of which rely on RINGs for their organization.
The science of RINGs is much like another Ring saga, full of peril and the unexpected. Unlike
that fellowship of the Ring, we continue our search for the "One Ring to rule them all, One Ring
to find them, One Ring to bring them all and in the darkness bind them" [131].
ACKNOWLEDGEMENTS
We thank Frank Rauscher and Jon Licht for access to work before publication. K.L.B.B.
acknowledges financial support from the NIH/NCI ROl CA80728-01.
REFERENCES
[I] Clarke, N. D., and Berg, J. M. (1998) Science 282, 2018-22.
[2] Freemont, P. S. (1993) Ann N Y Acad Sci 684, 174-92.
[3] Borden, K. L.B., and Freemont, P. S. (1996) Curr Opin Struct Biol 6, 395-401.
[4] Saurin, A. J., Borden, K. L.B., Boddy, 3VL N., and Freemont, P. S. (1996) Trends Biochem Sci 21, 208-14.
[5] Borden, K. L.B. (1998) Biochem Cell Biol 76, 351-8.
[6] Borden, K. L.B., Lally, J. M., Martin, S. R., O'Reilly, N. J., Etkin, L. D., and Freemont, P. S. (1995) EMBO J
14, 5947-56.
[7] Barlow, P. N., Luisi, B., Milner, A., Elliott, M., and Everett, R. (1994) JMol Biol 237, 201-11.
[8] Bellon, S. F., Rodgers, K. K., Schatz, D. G. ? Coleman, J. E., and Steitz, T. A. (1997) Nat Struct Biol 4, 586-91.
[9] Borden, K. L.B., Campbell Dwyer, E. J., Carlile, G. W., Djavani, M., and Salvato, M. S. (1998) J Virol 72, 3819-
26.
[10] Flores, T. P., Orengo, C. A., Moss, D. S., and Thornton, J. M. (1993) Protein Sci 2, 1811-26.
[II] Borden, K. L.B., Boddy, M. N., Lally, J., O'Reilly, N. J., Martin, S., Howe, K., Solomon, E., and Freemont, P. S.
(1995) EMBO J 14, 1532-41.
[12] Roehm, P. C, and Berg, J. M. (1997) Biochemistry 36, 10240-5.
[13] Miki, Y., Swensen, J., Shattuck-Eidens, D., Futreal, P. A., Harshman, K,, Tavtigian, S., Liu, Q., Cochran, C,
Bennett, L. M., Ding, W., and et al (1994) Science 266, 66-71.
[141 Brzovic, P. S., Meza, J., King, M. C, and Klevit, R. E. (1998) J Biol Chem 273, 7795-9.
[15] Wu, L. C, Wang, Z. W., Tsan, J. T., Spillman, M. A., Phung, A., Xu, X. L., Yang, M. C, Hwang, L. Y„
Bowcock, A. M., and Baer, R. (1996) Nat Genet 14, 430-40.
[16] Scully, R., Chen, J., Ochs, R. L., Keegan, K., Hoekstra, M., Feunteun, J., and Livingston, D. M. (1997) Cell 90,
425-35.
[17] Meza, J. E., Brzovic, P. S., King, M. C, and Klevit, R. E. (1999) J Biol Chem 274, 5659-65.
[18] Rodgers, K. K„ Bu, Z., Fleming, K. G., Schatz, D. G., Engelman, D. M., and Coleman, J. E. (1996) J Mol Biol
260, 70-84.
[19] Luisi, B. F., Xu, W. X., Otwinowski, Z., Freedman, L. P., Yamamoto, K. R., and Sigler, P. B. (1991) Nature 352,
497-505.
http://www.bentham.org/cpps 1 - l/L.B.Border^orden__c.htm
4/19/02
Construction of Macromolecular Assemblages in Eukaryotic Processes and their Role i... Page 30 of 33
[20
[21
[22
[23
[24
[25
[26
[27
[28
[29
[30
[31
[32
[33
[34
[35
[36
[37
[38
[39
[40
[41
[42
[43
[44
[45
[46
[47
[48
[49
[50
[51
[52
von Arnim, A. G., and Deng, X. W\ (1993) J Biol Chem 268, 19626-31.
Hill, A. V. (1910) J Physiol (London) 40, iv,
Takahashi, R„ Deveraux, Q., Tamm, L, Welsh, K., Assa-Munt, N., Saivesen, G. S., and Reed, J. C. (1998) J Biol
Chem 273, 7787-90.
Clem, R. J., and Miller, L. K. (1994) Mol Cell Biol 14, 5212-22.
van der Reijden, B. A., Erpelinck-Verscfaueren, C. A., Lowenberg, B., and Jansen, J. H. (1999) Protein Sci 8,
1557-61.
Morett, E., and Bork, P. (1999) Trends Biochem Sci 24, 229-31.
Torok, M., Freemont, P., and Etkin, L. D. (in press) Bioessays .
Borden, K. L., Lally, J. M„ Martin, S. R., O'Reilly, N. J., Solomon, E., and Freemont, P. S. (1996) Proc Natl
Acad Sci USA 93, 1601-6.
Peng, H., Begg, G. E., Schultz, D. C, Friedman, J. R., Jensen, D. E., Speicher, D. W., and Rauscher, F. Jr.
(2000) J. Mol. BioL 295, 1139-62.
Dhalluin, C, Carlson, J. E., Zeng, L., He, C, Aggarwal, A. K., and Zhou, M. M. (1999) Nature 399, 491-6.
Jack, L. J., and Mather, L H. (1990) J Biol Chem 265, 14481-6.
Tokunaga, C, Tatematsu, K., Kuroda, S., Nakagawa, N., and Kikkawa, U. (1998) FEBS Lett 435, 11-5.
Slack, F. J., and Ruvkun, G. (1998) Trends Biochem Sci 23, 474-5.
Jensen, D. E., Proctor, M., Marquis, S. T., Gardner, H. P., Ha, S. L, Chodosh, L. A., Ishov, A. M., Tommerup,
N., Vissing, H., Sekido, Y., Minna, J., Borodovsky, A., Schultz, D. C, Wilkinson, K. D., Maul, G. G., Barlev, N.,
Berger, S. L., Prendergast, G. C., and Rauscher, F. J., 3rd (1998) Oncogene 16, 1097-112.
Thompson, J. D., Higgins, D. G., and Gibson, T. J. (1994) Nucleic Acids Res 22, 4673-80.
Patarca, R., Freeman, G. J., Schwartz, J., Singh, R. P., Kong, Q. T., Murphy, E., Anderson, Y., Sheng, F. Y.,
Singh, P., Johnson, K. A., and et aL (1988) Proc Natl Acad Sci USA 55,2733-7.
Dadgostar, H., and Cheng, G. (1998) J Biol Chem 273, 24775-80.
Brink, R., and Lodish, H. F. (1998) J Biol Chem 273, 4129-34.
Callebaut, I., Labesse, G., Durand, P., Poupon, A., Canard, L., Chonulier, J., Henrissat, B., and Mornon, J. P.
(1997) Cell Mol Life Sci 53, 621-45.
Inouye, C, Dhillon, N., and Thorner, J. (1997) Science 278, 103-6.
Copps, K., Richman, R., Lyman, L. M., Chang, K. A., Ramper sad- Amnions, J., and Kuroda, M. L (1998)
l^MffO .7/7, 5409-17.
Topcu, Z., Mack, D. L., Hromas, R. A., and Borden, K. L. B, (1999) Oncogene 18, 7091-7100.
Houvras, Y. (1999) , Mt Sinai School of Medicine, New York.
Melnick, A., and Licht, J. D. (1999) Blood 93, 3167-215.
Wyman, J., and Gill, S. J. (1980) Proc Natl Acad Sci U S A 77, 5239-5242.
Fink, A. L. (1998) Fold Des 3, R9-23.
Bluemke, D. A., Carragher, B., Potel, M. J., and Josephs, R. (1987) Prog Clin Biol Res 240, 31-46.
Glover, J. R., Kowal, A. S„ Schirmer, E. C, Patino, M. M., Liu, J. J., and Lindquist, S. (1997) Cell 89, 811-9.
Uversky, V, N., Karnoup, A. S., Khurana, R., Segel, D. J., Doniach, S., and Fink, A. L. (1999) Protein Sci 8, 161-
73.
Freeh, C, and Schmid, F. X. (1995) J Mol Biol 251, 135-49.
Creighton, T. E. (1988) BioEssays 8, 57-63.
Creighton, T. E, (1984) Methods Enzymol 107, 305-329.
Eigen, M. (1963) Pure Appl Chem 6, 105.
http://www.bentham.org/cpps 1 - 1/L.B .Borden7borden_c.htm
4/19/02
Construction of Macromolecular Assemblages in Eukaryotic Processes and their Role i... Page 31 of 33
[53] Qi, P. X., Sosnick, TYR., and Englander, S. W. (1998) Nat Struct Biol 5, 882-4.
[54] Pearson, R. G. (1969) in Survey in Progress in Chemistry (Scott, A., Ed.), Vol. /, pp. 1, Academic Press, New
York.
[55] Eiser, B„ Kriz, W„ Bonventre, J. V., Englert, C, and Witzgall, R. (1997) J Biol Chem 272, 27908-12.
[56] Green, S., Kumar, V., Theulaz, L, Wahli, W., and Chambon, P. (1988) EMBO J 7, 3037-44.
[57] Rastelli, L., Chan, C. S., and Pirrotta, V. (1993) EMBO J 12, 1513-22.
[58] Kanno, M., Hasegawa, M., Ishida, A., Isono, K., and Taniguchi, M. (1995) EMBO J 14, 5672-8.
[59] Tanimura, S., Ohtsuka, S., Mitsui, K., Shirouzu, K., Yoshimura, A., and Ohtsubo, M. (1999) FEES Lett 447, 5-
9.
[60] Elenbaas, B., Dobbelstein, M., Roth, J., Shenk, T., and Levine, A. J. (1996) Mol Med 2, 439-51.
[61] Lai, Z., Freedman, D. A., Levine, A. J., and McLendon, G. L. (1998) Biochemistry 37, 17005-15.
[62] Marechal, V., Elenbaas, B., Piette, J., Nicolas, J. C, and Levine, A. J. (1994) Mol Cell Biol 14, 7414-20.
[63] Boddy, M. N., Freemont, P. S., and Borden, K. L. B. (1994) Trends Biochem Sci 19, 198-9.
[64] Borden, K. L. B., Campbell Dwyer, E. J., and Salvato, M. S. (1997) FEES Lett 418, 30-4.
[65] Mu, Z. M., Chin, K. V., Liu, J. H., Lozano, G., and Chang, K. S. (1994) Mol Cell Biol 14, 6858-67.
[66] Lai, H.-K., and Borden, K. L. B. (2000) Oncogene 19, 1623-1634 .
[67] Boddy, M. N„ Duprez, E., Borden, K. L. B., and Freemont, P. S. (1997) / Cell Sci 110, 2197-205.
[68] Cao, T., Duprez, E., Borden, K. L., Freemont, P. S., and Etkin, L. D. (1998) J Cell Sci 111, 1319-29.
[69] Le Douarin, B., Nielsen, A. L., You, J., Chambon, P., and Losson, R. (1997) Biochem Soc Trans 25, 605-12.
[70] Borden, K. L.B., Campbell Dwyer, E. J., and Salvato, M. S. (1998) J Virol 72, 758-66.
[71] Campbell Dwyer, E. J., Lai, H. K., MacDonald, R. C, Salvato, M. S., and Borden, K. L. B. (2000) J. Virol 74,
3293-3300.
[72] Chen, Y., Chen, C. F., Riley, D. J., Allred, D. C, Chen, P. L., Von Hoff, D„ Osborne, C. K., and Lee, W. H.
(1995) Science 270, 789-91.
[73] Scully, R., Ganesan, S., Brown, M., De Caprio, J. A., Cannistra, S. A., Feunteun, J., Schnitt, S., and Livingston,
D. M. (1996) Science 272, 123-6.
[74] Scully, R., Anderson, S. F., Chao, D. M„ Wei, W., Ye, L., Young, R. A., Livingston, D. 3VL, and Parvin, J. D.
(1997) Proc Natl Acad Sci USA 94, 5605-10.
[75] Chapman, M. S., and Verma, L M. (1996) Nature 382, 678-9.
[76] Mahajan, R., Delphin, C, Guan, T., Gerace, L., and Melchior, F. (1997) Cell 88, 97-107.
[77] Lucchesi, J. C. (1996) Bioessays 18, 541-7,
[78] Lyman, L. M., Copps, K., Rastelli, L., Kelley, R. L., and Kuroda, M. L (1997) Genetics 147, 1743-53.
[79] Palmer, M. J., Mergner, V. A., Richman, R., Manning, J. E., Kuroda, M. I., and Lucchesi, J. C. (1993) Genetics
134, 545-57.
[80] Vissing, H., Meyer, W. K., Aagaard, L., Tommerup, N., and Thiesen, H. J. (1995) FEES Lett 369, 153-7.
[81] Friedman, J. R., Fredericks, W. J., Jensen, D. E., Speicher, D. W., Huang, X. P., Neilson, E. G., and Rauscher,
F. J., 3rd (1996) Genes Dev 10, 2067-78.
[82] Moosmann, P., Georgiev, O., Le Douarin, B., Bourquin, J. P., and Schaffner, W. (1996) Nucleic Acids Res 24,
4859-67.
[83] Agata, Y., Matsuda, E., and Shimizu, A. (1999) J Biol Chem 274,
[84] Cryderman, D. E., Cuaycong, M. H., Elgin, S. C, and Wallrath, L. L. (1998) Chromosoma 107, 277-85.
[85] Ryan, R. F., Schultz, D. C, Ayyanathan, K., Singh, P. B., Friedman, J. R., Fredericks, W. J., and Rauscher, F.
J., 3rd (1999) Mol Cell Biol 19, 4366-78.
http://www.bentham.org/cppsl-l/L33orden/borden_c.htm
4/19/02
Constriction o'fMacromolecular Assemblages in HuKaryotic Processes and thefrHole Page 32 of 33
[861 Bach, L, Rodriguez-Esteban, C, Carriere, C, Bhushan, A., Krones, A., Rose, D. W., Glass, C. K., Andersen, B.,
Izpisua Belmonte, J. C, and Rosenfeld, M. G. (1999) Nat Genet 22, 394-9.
[87] Pirrotta, V. (1995) Curr Opin Genet Dev 5, 466-72.
[88] Hemenway, C. S„ Halligan, B. W., and Levy, L. S. (1998) Oncogene 16, 2541-7.
[89] Alkema, M. J., Bronk, ML, Verhoeven, E., Otte, A., van't Veer, L. J., Berns, A., and van Lohuizen, M. (1997)
Genes Dev 11, 226-40.
[90] Bunker, C. A., and Kingston, R. E. (1994) Mol Cell Biol 14, 1721-32.
[91] Bienz, M., and Muller, J. (1995) Bioessays 1 7, 775-84.
[92] Tyers, M., and Willems, A. R. (1999) Science 284, 601, 603-4.
[93] Phillips, N. C, and Dahman, J. (1995) Immunol Lett 45, 149-52.
(94] Seol, J. H., Feldman, R. M., Zachariae, W., Shevchenko, A., Correli, C. C, Lyapina, S., Chi, Y., Galova, M.,
Claypool, J., Sandmeyer, S., Nasmyth, K., and Deshaies, R. J. (1999) Genes Dev 13, 1614-26.
[95] Skowyra, D., Koepp, D. M., Kamura, T., Conrad, M. N., Conaway, R. C, Conaway, J. W., Elledge, S. J., and
Harper, J. W. (1999) Science 284, 662-5.
[96] Ohta, T., Michel, J. J., Schottelius, A. J., and Xiong, Y. (1999) Mol Cell 3, 535-41.
[97] Kwon, Y. T., Levy, F., and Varshavsky, A. (1999) J Biol Chem 274, 18135-9.
[98] Kwon, Y. T., Reiss, Y„ Fried, V. A., Hershko, A., Yoon, J. K., Gonda, D. K., Sangan, P., Copeland, N. G.,
Jenkins, N. A., and Varshavsky, A. (1998) Proc Natl Acad Sci U S A 95, 7898-903.
[99] Honda, R., Tanaka, H., and Yasuda, H. (1997) FEBS Lett 420, 25-7.
[100] Waterman, H., Levkowitz, G., Alroy, L, and Yarden, Y. (1999) / Biol Chem 274, 22151-4.
[101] Martinez-Noel, G., Niedenthal, R., Tamura, T., and Harbers, K. (1999) FEBS Lett 454, 257-61.
[1021 Bordallo, J., Plemper, R. K., Finger, A., and Wolf, D. H. (1998) Mol Biol Cell 9, 209-22.
[103] Nakayama, M., Miyake, T., Gahara, Y., Ohara, O., and Kitamura, T. (1995) / Neurosci 15, 5238-48.
[104) Lorick, K. L., Jensen, J. P., Fang, S., Ong, A. M., Hatakeyama, S., and Weissman, A. M. (1999) Proc Natl Acad
Sci US A 96, 11364-9.
[105] Joazeiro, C. A., Wing, S. S., Huang, H., Leverson, J. D., Hunter, T., and Liu, Y. C. (1999) Science 286, 309-312.
[1061 Matunis, M. J., Wu, J., and Blobel, G. (1998) J Cell Biol 140, 499-509.
[107] Wajant, H., Muhlenbeck, F., and Scheurich, P. (1998) J Mol Evol 47, 656-62.
[1081 Takeuchi, M., Rothe, M., and Goeddel, D. V. (1996) J Biol Chem 271, 19935-42.
[109] Sato, T., Irie, S., and Reed, J. C. (1995) FEBS Lett 358, 113-8.
[110] Cheng, G., Cleary, A. M., Ye, Z. S., Hong, D. I., Lederman, S„ and Baltimore, D. (1995) Science 267, 1494-8.
[Ill] Rothe, M., Sarma, V., Dixit, V. M., and Goeddel, D. V. (1995) Science 269, 1424-7.
[112] Ishida, T. K., Tojo, T., Aoki, T., Kobayashi, N., Ohishi, T., Watanabe, T., Yamamoto, T., and Inoue, J. (1996)
Proc Natl Acad Sci U S A 93, 9437-42.
[113] Baud, V., Liu, Z. G., Bennett, B., Suzuki, N., Xia, Y., and Karin, M. (1999) Genes Dev 13, 1297-308.
[114] Scotland, P. B., Colledge, M., Melnikova, L, Dai, Z„ and Froehner, S. C. (1993) / Cell Biol 123, 719-28.
[115] Rothe, M., Pan, M. G., Henzel, W. J., Ayres, T. M., and Goeddel, D. V. (1995) Cell 83, 1243-52.
[1161 Matsumoto, M., Lo, S. F., Carruthers, C. J., Min, J., Mariathasan, S., Huang, G., Plas, D. R., Martin, S. M.,
Geha, R. S., Nahm, M. H., and Chaplin, D. D. (1996) Nature 382, 462-6.
[117] Nakano, H., Oshima, H., Chung, W., Williams-Abbott, L., Ware, C. F„ Yagita, H., and Okumura, K. (1996) J
Biol Chem 271, 14661-4.
[118] Burd, C. G., Peterson, M., Cowles, C. R., and Emr, S. D. (1997) Mol Biol Cell 8, 1089-104.
[119[ Vitale, N., Moss, J., and Vaughan, M. (1996) Proc Natl Acad Sci U S A 93, 1941-4.
http://www.bentham.org/cpps 1 - 1/L.B ,Border^orden_c.htm
4/19/02
/Constru'ctiorfof Macromolecular Assemblages in Eukaryotic Processes andtlieir Role i... Page 33 of 33
[1201 Patarca, R., and Fletcher, M. A. (1992) FEES Lett 312, 1-2.
[121] Tsukamoto, T., Shimozawa, N., and Fujiki, Y. (1994) Mol Cell Biol 14, 5458-65.
[122] Titorenko, V. I., Waterham, H. R„ Cregg, J. M., Harder, W., and Veenhuis, M. (1993) Proc Natl Acad Sci USA
90, 7470-4.
[123] Kalish, J, E., Chen, C. I., Gould, S. J., and Watkins, P. A. (1995) Biochem Biophys Res Commun 206, 335-40.
[124] Kalish, J. E., Theda, C, Morrell, J. C., Berg, J. M., and Gould, S. J. (1995) Mol Cell Biol 15, 6406-19.
[125] El-Husseini, A. E., and Vincent, S. R. (1999) J Biol Chem 274, 19771-7.
[126] Thien, C. B., Bowtell, D. D., and Langdon, W. Y. (1999) J Immunol 162, 7133-9.
[127] Blake, T. J., Shapiro, M., Morse, H. C. d„ and Langdon, W. Y. (1991) Oncogene 6, 653-7.
[128] Langdon, W. Y., Hartley, J. W., Klinken, S. P., Ruscetti, S. K., and Morse, H. C. d. (1989) Proc Natl Acad Sci U
SA 86, 1168-72.
[129] Levkowitz, G., Waterman, H., Zamir, E., Kam, Z., Oved, S., Langdon, W. Y., Beguinot, L„ Geiger, B., and
Yarden, Y. (1998) Genes Dev 12, 3663-74.
[130] Miyake, S., Lupher, M. L., Jr., Druker, B., and Band, H. (1998) Proc Natl Acad Sci USA 95, 7927-32.
[131] Tolkien, J. R. R. (1954) The Lord of the Rings, Ballantine Books, New York.
[132] Dawid, I. B., Breen, J. J., and Toyama, R. (1998) Trends Genet 14, 156-62.
[133] Henry, J., Ribouchon, M. T., Offer, C, and Pontarotti, P. (1997) Biochem Biophys Res Commun 235, 162-5.
[134] Borden, K. L. B. (2000) /. Mol Biol 295, 7091-7100.
http://www.benth^^
4/19/02